Research, Society and Development, v. 15, n. 10, e35151051799, 2026 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v15i10.51799 Familiopathic and thymopathic borderline: Proposal of an etiological typology of borderline personality disorder Borderline familiopático e borderline timopático: Proposta de uma tipologia etiológica do transtorno de personalidade borderline Borderline familiopático y borderline timopático: Propuesta de una tipología etiológica del trastorno límite de la personalidad Received: 10/02/2026 | Accepted: 10/07/2026 | Published: 10/09/2026 Giovanna Azevedo Rodrigues ORCID: https://orcid.org/0009-0003-9020-9368 ASMIGO Psychiatric Hospital, Brazil E-mail: giovannagirodrigues@hotmail.com Marcelo Caixeta ORCID: https://orcid.org/0000-0003-2357-6558 ASMIGO Psychiatric Hospital, Brazil Federal University of Goiás, Brazil E-mail: cfm11041963@ufg.br Abstract Borderline personality disorder (BPD) is clinically heterogeneous, which helps explain the uneven response of patients to treatment. This article aims to propose an etiological typology that sorts patients by the relative weight of two sources of instability, the family and mood. This theoretical essay is supported by a directed narrative review of 30 references, mostly register-based, genetic and meta-analytic studies and guidelines from 2021 to 2026. In familiopathic borderline, instability is maintained by a pathogenic family environment still active; medication acts only on symptoms, and the decisive intervention targets the family and the protection of the patient. In thymopathic borderline, instability expresses a constitutional affective dysregulation, with a hypothesized delayed response to lithium. The mixed form combines both and is argued to predominate in Brazilian practice. Recent findings support both axes and the rarity of the pure familial form: heritability of 46%, a high genetic correlation with depression and a moderate one with bipolar disorder, and a minimal contribution of the environment shared by siblings. The axes interact, since part of the inherited liability appears to be a greater sensitivity to adversity, and family load is distinguished as active or past. Maltreatment by a nonparent maternal partner is proposed as the most destructive family factor. The pharmacological component exceeds current evidence and is presented as a hypothesis. A five-step classification algorithm, an evidence map, implications for early identification and four testable predictions are provided. The typology offers the clinician a simple question and a research agenda, pending empirical validation. Keywords: Borderline personality disorder; Etiology; Family relations; Mood disorders; Lithium. Resumo O transtorno de personalidade borderline (TPB) é clinicamente heterogêneo, o que ajuda a explicar a resposta desigual ao tratamento. O objetivo é propor uma tipologia etiológica que distingue os pacientes pelo peso de duas fontes de instabilidade, a família e o humor. Ensaio teórico apoiado em revisão narrativa dirigida de 30 referências, na maioria estudos de registro, genéticos e meta-analíticos e diretrizes de 2021 a 2026. No borderline familiopático, a instabilidade é mantida por um ambiente familiar patogênico ativo; o fármaco atua apenas sobre sintomas, e a intervenção decisiva recai sobre a família e a proteção do paciente. No borderline timopático, a instabilidade expressa uma desregulação afetiva constitucional, com hipótese de resposta tardia ao lítio. A forma mista combina os dois e predominaria na clínica brasileira. Achados recentes apoiam os dois eixos e a raridade da forma familiar pura: herdabilidade de 46%, correlação genética alta com depressão e moderada com transtorno bipolar, e contribuição mínima do ambiente comum aos irmãos. Os eixos interagem, pois parte da predisposição herdada parece ser uma sensibilidade maior à adversidade, e a carga familiar pode ser ativa ou passada. O maltrato por parceiro materno não genitor é proposto como o fator familiar mais destrutivo. O componente farmacológico excede a evidência atual e é apresentado como hipótese. Oferecem-se um algoritmo de classificação em cinco passos, um mapa de evidências, implicações para a identificação precoce e quatro previsões testáveis. A tipologia dá ao clínico uma pergunta simples e uma agenda de pesquisa, pendente de validação empírica. Palavras-chave: Transtorno da personalidade borderline; Etiologia; Relações familiares; Transtornos do humor; Lítio. 1 Research, Society and Development, v. 15, n. 10, e35151051799, 2026 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v15i10.51799 Resumen El trastorno límite de la personalidad (TLP) es clínicamente heterogéneo, lo que ayuda a explicar la respuesta desigual al tratamiento. El objetivo es proponer una tipología etiológica que distingue a los pacientes según el peso de dos fuentes de inestabilidad, la familia y el humor. Ensayo teórico apoyado en revisión narrativa dirigida de 30 referencias, sobre todo estudios de registro, genéticos y metaanalíticos de 2021 a 2026. En el borderline familiopático, un ambiente familiar patógeno activo mantiene la inestabilidad; el fármaco actúa solo sobre síntomas y la intervención decisiva recae en la familia y la protección del paciente. En el borderline timopático, la inestabilidad expresa una desregulación afectiva constitucional, con hipótesis de respuesta tardía al litio. La forma mixta combina ambos y predominaría en la clínica brasileña. Hallazgos recientes apoyan ambos ejes y la rareza de la forma familiar pura: heredabilidad del 46%, correlación genética alta con depresión y moderada con trastorno bipolar, y contribución mínima del ambiente compartido. Los ejes interactúan, pues parte de lo heredado parece ser una mayor sensibilidad a la adversidad, y la carga familiar puede ser activa o pasada. El maltrato por la pareja materna no progenitora sería el factor familiar más destructivo. El componente farmacológico excede la evidencia actual y se presenta como hipótesis. Se ofrecen un algoritmo de clasificación en cinco pasos, un mapa de evidencias, implicaciones para la identificación precoz y cuatro predicciones comprobables. La tipología da al clínico una pregunta sencilla y una agenda de investigación, pendiente de validación empírica. Palabras clave: Trastorno de personalidad limítrofe; Etiología; Relaciones familiares; Trastornos del humor; Litio. 1. Introduction Borderline Personality Disorder (BPD) brings together, under one diagnostic label, patients who became unstable by different routes. Estimates of its prevalence in the general population range from 0.7% to 7.35%, and people with the diagnosis account for a large share of contacts with psychiatric services (Ramos-Suárez et al., 2026; Hall et al., 2026). The disorder combines affective dysregulation, impulsivity, identity disturbance and unstable relationships, but the proportion of each varies widely from one patient to the next (Leichsenring et al., 2024). Clinicians meet this variation every day in the form of an unpredictable response to treatment: a regimen that settles one patient leaves the course of another untouched. Most attempts to make sense of this heterogeneity have been descriptive. A systematic review of cluster-analytic studies identified symptom-based subtypes that recur across samples, yet it also found inconsistent methods and almost no longitudinal or treatment-outcome data (Triantafyllou et al., 2025). A subtype defined by its symptom profile describes how a patient is unstable. It does not say why, and the clinician who must choose an intervention needs the second answer more than the first. Research on causes points in two directions at once. The heritability of clinically diagnosed BPD was estimated at 46% in a Swedish register study, and the first genome-wide study to find significant loci confirmed that the disorder has a polygenic basis (Skoglund et al., 2021; Streit et al., 2026). At the same time, 71.4% of patients report at least one childhood adversity, and the odds of such a report are 13.91 times those of non-clinical controls (Porter et al., 2020; Kleindienst et al., 2021). Neither line of evidence accounts for every case. Close to 30% of patients report no adversity at all, and most maltreated children never develop the disorder. The proposal presented here grew out of two recurring clinical observations. Some patients improve markedly when they are away from their families and worsen when they return, whatever medication they are taking. Others change little with a change of setting but stabilize some months after a mood stabilizer is introduced. The two observations suggest that instability can be maintained from outside the patient or from within, and the typology is an attempt to state that difference in a form that can be examined. These two bodies of evidence are usually read as competing explanations or folded into a single interactive model. The proposal developed here takes a different step and treats them as two sources of instability whose relative weight differs between patients. This article aims to propose an etiological typology that sorts patients by the relative weight of two sources of instability, the family and mood. The purpose is practical: to help decide, for each patient, where the intervention capable of changing the course is to be found. 2 Research, Society and Development, v. 15, n. 10, e35151051799, 2026 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v15i10.51799 2. Methodology This is a theoretical essay (Santos de Souza, 2025; Risemberg et al., 2026a) of qualitative nature (Risemberg et al., 2026b; Pereira et al., 2018), supported by a directed, non-systematic narrative review (Ogassawara et al., 2025; Rother, 2007). The typology originated in clinical observation of patients with BPD and was then confronted with the literature in a search carried out in October 2026. The search gave priority to register-based population studies, genetic studies, meta-analyses, Cochrane reviews and practice guidelines published between 2021 and 2026. Older works were retained only when they are the primary source of a concept or the only available trial on a given question. The final set comprises 30 references, 23 of which were published from 2021 onwards. Table 1 shows how they are distributed by category of evidence. Table 1 - Distribution of the 30 references by category of evidence. Category n References Genetic, twin and family register studies 5 Skoglund et al. (2021); Skaug et al. (2022); Fan et al. (2025); Hall et al. (2026); Streit et al. (2026) Meta-analyses and systematic reviews: etiology, epidemiology and classification 6 Porter et al. (2020); Kleindienst et al. (2021); Durdurak et al. (2022); Lee et al. (2022); Triantafyllou et al. (2025); Ramos-Suárez et al. (2026) Meta-analyses and systematic reviews: treatment 6 Storebø et al. (2020); Gartlehner et al. (2021); Guillén et al. (2021); Stoffers-Winterling et al. (2022); Pereira Ribeiro et al. (2025); Gerolymos et al. (2026) Clinical trial 1 Links et al. (1990) Practice guideline 1 Keepers et al. (2024) Epidemiology of child maltreatment 3 Schnitzer & Ewigman (2005); Wilson et al. (2023); Malta et al. (2025) Narrative reviews and theoretical works 8 Linehan (1993); Akiskal (2004); Paris & Black (2015); Bozzatello et al. (2021); Saccaro et al. (2021); Almeida & Sanches (2024); Leichsenring et al. (2024); Gallicchio et al. (2026) Source: Authors (2026). As Table 1 shows, the etiological part of the argument rests mainly on genetic studies and meta-analyses, whereas the therapeutic part depends on syntheses of trials and, for lithium, on a single small trial. To make this imbalance explicit, each thesis of the model was graded according to the support found in the literature (strong, moderate, weak or absent), and the result is presented as an evidence map in Section 3.7. No patient data were collected, so ethical review was not required. 3. Results and Discussion 3.1 Empirical foundations Four groups of findings set the limits within which any etiological typology of BPD has to work: the genetic architecture of the disorder, the nature of what is inherited, the contribution of the family environment, and the unsettled boundary with the bipolar spectrum. 3.1.1 Genetic architecture The Swedish register study followed more than 1.8 million individuals, 11,665 of whom received a diagnosis of BPD. The risk of the disorder rose with genetic proximity to an affected relative: the hazard ratio was 4.7 for full siblings and 11.5 for monozygotic twins. Heritability was estimated at 46%, and the remaining variance was attributed to environmental factors unique to the individual (Skoglund et al., 2021). Molecular data now complement these figures. The largest genome-wide association study to date, with 12,339 cases and more than one million controls, identified 11 independent risk loci and estimated the heritability attributable to common variants at 17.3%. A polygenic score derived from these results explained 4.6% of the variance in liability (Streit et al., 2026). 3 Research, Society and Development, v. 15, n. 10, e35151051799, 2026 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v15i10.51799 The numbers are modest, as they are for most psychiatric disorders, but they settle one point: BPD has a measurable constitutional component, and an etiological model without a constitutional term is incomplete. 3.1.2 The nature of the inherited liability What are inherited matters as much as how much. In the genome-wide study, the strongest genetic correlations of BPD were with post-traumatic stress disorder, depression, attention-deficit/hyperactivity disorder, antisocial behavior and measures of suicide and self-harm. The correlation with bipolar disorder was lower, at 0.35, although still substantial (Streit et al., 2026). The Swedish family study of 2.7 million individuals reached a compatible result by a different method: the genetic correlation between BPD and depressive disorder was 0.92, and the correlation with self-harm was 0.80 (Hall et al., 2026). Two conclusions follow. The liability transmitted in the families of patients with BPD is largely affective, which gives biological content to the idea of a mood-based route to the disorder. That liability, however, lies closer to depression and to reactivity to stress than to classic bipolarity. The overlap with bipolar disorder is real, and it extends to the inflammatory and stress-response alterations described in both conditions (Saccaro et al., 2021), but it is partial. 3.1.3 Family environment and genetic confounding Family studies also constrain what can be claimed about the home. In the Swedish family study, the associations between BPD and other psychiatric disorders were explained in roughly equal parts by genetic factors and by environmental factors unique to the individual, while the environment shared by siblings contributed between 0% and 14% (Hall et al., 2026). A model in which a harmful household produces BPD in whoever grows up in it is therefore hard to sustain. Three details qualify this conclusion and keep the family in the picture. First, maltreatment aimed at one particular child counts, in the logic of these studies, as non-shared environment, because it does not make siblings alike. Second, the largest shared-environment contribution, 14%, was found for the association between BPD and post-traumatic stress disorder, which is the association most directly tied to events that a household can impose on its members (Hall et al., 2026). Third, in both Swedish studies the familial risk was somewhat higher for maternal than for paternal half-siblings, 2.1 against 1.3 in the earlier analysis (Skoglund et al., 2021; Hall et al., 2026). Half-siblings of both kinds share the same proportion of genes, but maternal halfsiblings more often grow up in the same home, so the difference suggests that the household leaves a trace, even if a small one. The category of non-shared environment needs to be read with some caution as well. It includes everything that makes siblings differ, from distinct experiences to error of measurement (Hall et al., 2026), and its size is not a precise estimate of the effect of individual experience. What the family studies establish is narrower and more solid: growing up in the same household does not, by that fact alone, make siblings alike in their risk of BPD. A further difficulty is that adversity itself is not independent of genes. In a discordant twin study, the correlation between childhood trauma and borderline traits was mostly genetic (0.53), with a correlation of only 0.11 for environmental factors unique to the individual (Skaug et al., 2022). The Swedish registers show the same pattern for assault and victimization, whose genetic correlation with BPD was 0.68 (Hall et al., 2026). The genetic load that predisposes to the disorder also raises the probability of being exposed to adverse environments. Assortative mating reinforces the effect: spousal correlations for psychiatric disorders are consistent across cultures and persist over generations (Fan et al., 2025), so a child may inherit liability from both parents and be raised in a household shaped by the difficulties of both. 3.1.4 Adversity and invalidation None of this cancels the weight of adversity. In the meta-analysis by Porter et al. (2020), which pooled 97 studies, 4 Research, Society and Development, v. 15, n. 10, e35151051799, 2026 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v15i10.51799 emotional abuse (OR 38.11) and neglect (OR 17.73) showed the strongest associations with BPD, and patients with BPD were 3.15 times more likely to report adversity than patients with other psychiatric disorders. The association is specific as well as strong. Its size, on the other hand, depends on the design: in cohort and epidemiological studies the effect falls to about 2.6, a sign that retrospective report in clinical samples inflates the estimate. The ranking of adversities in that meta-analysis is instructive. The strongest associations were with emotional abuse and neglect, forms of adversity that leave no physical mark and seldom reach a notification system. What harms, on this evidence, is less the isolated violent event than the daily climate in which the emotional life of the child is denied or attacked. Adversity need not take the form of overt abuse. A multivariate meta-analysis of 21 studies with 7,198 participants found that parental invalidation was associated with borderline symptoms with an effect of 0.25. Maternal invalidation carried a slightly larger effect than paternal invalidation, and neither age nor culture changed the association (Lee et al., 2022). Reviews of early risk factors add parental psychopathology and the temperament of the child to this list (Bozzatello et al., 2021). The biosocial model is the framework that holds these findings together. It describes BPD as the product of a transaction between a constitutional emotional vulnerability and an environment that invalidates the emotional experience of the child (Linehan, 1993). The typology proposed here accepts this framework and asks a question that the framework leaves open: which of the two terms carries more weight in each patient. 3.1.5 The boundary with the bipolar spectrum The place of mood in BPD has been disputed for decades. Akiskal (2004) argued that borderline personality belongs to the bipolar spectrum. Paris and Black (2015) defended the opposite view, that the two are distinct conditions, and placed the difference in the temporal pattern of affect: in BPD, mood shifts are brief and reactive to interpersonal stressors, whereas in bipolar disorder mood changes are sustained. They also warned that mistaking one for the other leads to treatment that does not help. A meta-review of 19 systematic reviews, covering 217 prospective studies, did not settle the matter. It found a long list of predisposing factors common to both disorders, among them family history of psychopathology, affective instability, sleep disturbance and childhood adversity, and concluded that the available data neither establish two separate entities nor justify treating BPD as an extension of bipolar disorder (Durdurak et al., 2022). The thymopathic type described below is in this disputed territory, and the genetic findings summarized above indicate where in that territory it should be placed. 3.1.6 Previous approaches to heterogeneity Research on the heterogeneity of BPD has followed two paths. One looks for the dimensions that underlie the diagnostic criteria by means of factor analysis. The other looks for groups of patients by means of cluster-analytic methods such as latent class and latent profile analysis. A systematic review of the second path found subtype patterns that recur from study to study, but the studies differed so much in samples, measures and statistical procedures that the patterns could not be consolidated, and very few followed patients over time or related the subtypes to treatment outcome (Triantafyllou et al., 2025). Both paths start from the symptoms and work back towards groups. The typology proposed here goes the other way: it starts from two presumed causes and predicts the clinical pattern that each should produce. The advantage of this direction is that the predictions concern course and response to treatment, which are precisely the data that the descriptive subtypes lack. The cost is that the causes must be inferred from history and observation, with the uncertainty that this entails. 5 Research, Society and Development, v. 15, n. 10, e35151051799, 2026 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v15i10.51799 3.2 The proposed typology The model places each patient on two axes. The first, family load, measures how far a pathogenic family environment is still acting on the patient. The second, thymic load, measures the weight of a constitutional affective dysregulation. Figure 1 shows the four regions that result from combining them. Figure 1 - Two-axis model of the etiological typology of borderline personality disorder. Source: Authors (2026). In Figure 1, the two pure types occupy the corners in which only one load is high, and the mixed form occupies the region in which both are high. When neither load can be identified, the diagnosis itself deserves to be reviewed. The axes are continuous, so the types are better understood as regions with gradual borders than as separate categories, and a patient may move between them when the family situation changes. 3.2.1 Familiopathic borderline Familiopathic borderline is the presentation in which instability is generated and maintained by a pathogenic family environment that remains active. The term shifts the seat of the illness from the individual to the system in which the individual lives. The patient expresses the pathology of that system, and remission depends on change within it. The pathogenic environment takes several forms, which can be arranged by degree. The mildest, and probably the most common, is chronic invalidation, in which the emotions of the child are habitually dismissed, ridiculed or punished. Its association with borderline symptoms is of small to moderate size, slightly larger for the mother than for the father, and it holds across ages and cultures (Lee et al., 2022). Next come emotional abuse and neglect, the adversities with the strongest associations in the meta-analytic literature (Porter et al., 2020). At the far end is physical and sexual violence. In the Brazilian notification data, the aggressor in violence against young children was most often the mother or the father, and the place was the home in 88.3% of cases (Malta et al., 2025). The agent of harm in this type is therefore usually someone on whom the child depends, which is what makes the exposure both hard to escape and hard to report. The mechanism proposed is sustained invalidation. When the emotional expression of a child is punished, ignored or answered with violence, the child does not learn to name and modulate affect, and extreme behavior becomes the only signal 6 Research, Society and Development, v. 15, n. 10, e35151051799, 2026 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v15i10.51799 that the environment registers (Linehan, 1993; Lee et al., 2022). As long as the environment stays the same, this learning is rehearsed every day. The clinical picture is correspondingly reactive. Affective shifts last hours and follow identifiable relational events, most often a conflict, a humiliation or a threat of abandonment within the family. Anger and hostility are prominent, self-injury tends to occur in the hours after a confrontation, and transient suspiciousness or dissociative experiences may appear at the height of a crisis. The feature that clearly marks the type is its dependence on context. The same patient who is unmanageable at home may be calm at school, at work or in a ward, and symptoms abate during hospital admission only to return after discharge. Relatives often present the patient as the single problem of the household and are reluctant to examine their own part in it, and this reluctance is itself a clinical sign. History usually begins in childhood or adolescence, and the onset or worsening of symptoms can often be dated to a change in the household, such as separation, the arrival of a new adult or the illness of a parent. Maltreatment and parental psychopathology are among the early predictors of the disorder (Bozzatello et al., 2021). In adults, the active environment may be the family of origin with which the patient still lives, or on which the patient still depends, or a later relationship that reproduces the same pattern of invalidation and violence. The criterion is the same in each case: the harmful exposure has not ended. This explains the limited reach of medication in this type. No drug treats the core psychopathology of BPD (StoffersWinterling et al., 2022), and the agents with some support act on hostility and anger (Gerolymos et al., 2026). An antipsychotic can therefore reduce hostility, impulsivity and transient cognitive-perceptual symptoms while exposure continues, but the improvement does not hold, because the cause is still operating. A particular risk follows. Each relapse after a conflict at home is easily read as a failure of the drug, and the response is to raise the dose or to add another agent. The guideline of the American Psychiatric Association advises that medication be limited in time and aimed at a defined target symptom, and that current and past medications be reviewed before a new one is started (Keepers et al., 2024). In the familiopathic type this advice has a specific meaning: escalating treatment in the patient while the family remains unchanged adds adverse effects without touching the cause. The intervention with the power to change the course is the one that reaches the family. Programs designed for relatives, of which Family Connections is the best known, combine education about the disorder with training in how to respond to the patient, and they produced favorable results across different formats and settings in a systematic review (Guillén et al., 2021). Two reservations apply. The authors stressed the limited methodological quality of the studies, and the programs were designed chiefly to relieve the distress and burden of carers, so their effect on the course of the patient, which is what the model predicts, has yet to be demonstrated. Structured psychotherapy for the patient remains indicated (Storebø et al., 2020), but while the exposure continues its gains are fragile, and it works best when the family is engaged at the same time. Where there is violence, protection and removal come before any other measure. The study of fatal child abuse discussed in Section 3.3 offers two warning signs to the clinician: deaths precipitated by harsh punishment more often involved a child who had already suffered a non-fatal injury and an adult who knew of earlier abuse (Wilson et al., 2023). In a child or adolescent with borderline features, a previous injury and an adult who knows and does not act call for notification and for the involvement of protective services. The model predicts that the course of this type follows the course of the family. When the environment changes, through effective work with the relatives, the departure of an aggressor or the removal of the patient, reactive crises should diminish within weeks, and what remains is the learned pattern discussed in Section 3.2.5. When the environment does not change, no treatment directed at the patient alone is expected to hold. The pure form is rare. For it to occur, a child with no constitutional vulnerability must be exposed to severe adversity 7 Research, Society and Development, v. 15, n. 10, e35151051799, 2026 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v15i10.51799 to produce the condition by itself. The small contribution of shared environment in family studies and the genetic confounding of the link between trauma and borderline traits both indicate that this is the exception (Hall et al., 2026; Skaug et al., 2022). In practice, many cases that look familiopathic at first assessment will turn out to carry a thymic component as well. 3.2.2 Thymopathic borderline Thymopathic borderline, from the Greek thymos (mood), is the presentation in which instability expresses a constitutional affective dysregulation. The family may function well, and the symptoms persist when the patient is away from it. Constitutional is used here in a precise sense. The liability to BPD is heritable, with an estimate of 46% and a risk that rises with the degree of kinship to an affected relative (Skoglund et al., 2021). It is polygenic (Streit et al., 2026), and the part of it that is shared with other disorders is shared above all with depression (Hall et al., 2026). The thymopathic type is the clinical expression of this liability when it is strong enough to produce instability without the help of an adverse environment. The clinical picture differs from the familiopathic one in its relation to events. Alongside the usual reactivity, the history shows periods of several days in which mood, energy and sleep change without a proportionate trigger. Irritability is more common than elation in these periods and stretches of low mood and withdrawal alternate with stretches of tension and restlessness. Self-injury, when it occurs, tends to accompany the low periods and need not follow a conflict. Because the swings do not depend on context, they appear at home, at work and on the ward alike, and a change of setting makes little difference. There is often bipolar disorder or recurrent depression among first-degree relatives, and a cyclothymic or irritable temperament that can be traced back to childhood or adolescence, in keeping with the place of temperament among the early predictors of the disorder (Bozzatello et al., 2021). The relation of this type to bipolar disorder must be stated carefully. Akiskal (2004) saw in borderline personality a member of the bipolar spectrum, and the thymopathic type takes from that view the idea that some patients have an affective temperament at the root of their instability. It departs from it in two respects. The idea is applied to a subgroup and not to the diagnosis as a whole, and the genetic findings place the subgroup nearer to depression than to bipolar disorder, since the inherited load of BPD correlates strongly with the former and only moderately with the latter (Streit et al., 2026; Hall et al., 2026). In this model, therefore, thymopathy means a broad constitutional dysregulation of affect. The autonomous periods described above are longer than the reactive shifts that Paris and Black (2015) regard as typical of BPD, but they fall short of the sustained episodes that define bipolar disorder. A patient with a frank manic or hypomanic episode has comorbid bipolar disorder, which is a different matter and is treated as such. Depression deserves particular attention in this type. In the Swedish registers, depressive disorder was among the conditions most strongly associated with BPD in the same individual, and the genetic correlation between the two was 0.92 (Hall et al., 2026). The depressive symptoms that accompany BPD have been the object of a specific analysis of the trial literature (Pereira Ribeiro et al., 2025). In the thymopathic patient, depressive stretches are expected to be part of the autonomous oscillation, and their presence in a patient whose family environment is adequate strengthens the classification. Sleep is the most useful of these markers in practice. Sleep disturbance is among the predisposing factors common to BPD and bipolar disorder in prospective studies of young people (Durdurak et al., 2022), and in the Swedish family study sleep disorders showed the strongest genetic overlap with BPD among all the somatic conditions examined, with a genetic correlation of 0.74 (Hall et al., 2026). Changes in sleep that arise without an external reason are therefore a reasonable clinical sign of thymic load, and they are easier to document than changes in mood. The role of the family in this type is different, and the difference has practical weight. Living with a person with BPD places relatives under considerable strain (Guillén et al., 2021), and a family worn down by years of crises may look dysfunctional 8 Research, Society and Development, v. 15, n. 10, e35151051799, 2026 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v15i10.51799 at first contact. The disorder is then easily attributed to the family, when the family is bearing its consequences. This is the error of classification that the thymopathic type is meant to prevent. Relatives in this situation benefit from the programs of education and support developed for carers, and they are allies in recording the course of mood and sleep. The therapeutic hypothesis attached to this type is a response to lithium that appears late, after about six months, with an antipsychotic serving as a bridge during that interval. The direct evidence is thin. The only randomized trial was a crossover study comparing lithium, desipramine and placebo, in which seventeen patients were enrolled and ten completed at least two of the three arms. It found no difference on depression scales, although clinicians rated anger and suicidal symptoms as improved during lithium treatment (Links et al., 1990). Two recent reviews reach similar conclusions: lithium is a plausible option for impulsivity, aggression and suicidality in BPD, in part because of its effects on neuroplasticity and inflammation, but adequate trials are lacking (Almeida & Sanches, 2024; Gallicchio et al., 2026). No trial with a follow-up of six months was located, so the proposed latency remains a clinical observation that has yet to be tested. The model speaks of lithium, but the available trials offer indirect signals for other agents. In the network meta-analysis, lamotrigine and topiramate reduced hostility and anger in unselected samples, and aripiprazole did the same among the antipsychotics, whereas the evidence for haloperidol was of low certainty (Gerolymos et al., 2026). These results bear on the choice of the bridging drug, and they suggest that the question for future trials is broader than lithium: whether mood stabilizers as a class benefit patients selected for thymic load. The biological argument for lithium in this type is indirect. Inflammatory and stress-related alterations have been described in both bipolar disorder and BPD (Saccaro et al., 2021), and the anti-inflammatory and neuroplastic actions of lithium are among the reasons given for studying it in BPD (Gallicchio et al., 2026). The reasoning is plausible, but none of its links has been tested in patients selected by the criteria proposed here. What the model predicts as a response is also specific. The stabilizer is expected to reduce the amplitude and frequency of the autonomous swings over months. It is not expected to remove the sensitivity to rejection or the difficulties of identity and relationship, which are the object of structured psychotherapy (Storebø et al., 2020). A thymopathic patient who is well stabilized may thus continue to meet criteria for BPD, with a milder and more predictable course. Judging the treatment by the disappearance of the diagnosis would lead to the false conclusion that it had failed. One reading of the trial literature favors the model. If lithium helps only patients of the thymopathic type, a trial that recruits unselected patients with BPD will dilute the effect, and a null result is what one would expect. The same reasoning applies to other mood stabilizers. This is a prediction that can be tested, and it is taken up again in Section 3.9. Because the therapeutic consequence of this classification is a long-term drug with real risks, the threshold for assigning the type should be high. The model requires, at a minimum, autonomous oscillation documented prospectively and a family history of mood disorder. A reactive pattern alone, however intense, does not qualify. This restriction answers the warning of Paris and Black (2015) that reactive instability treated as a mood disorder leads to ineffective prescribing. 3.2.3 Mixed form In the mixed form, thymic vulnerability and family adversity are both present and each aggravates the other. The proposal is that this is the most common presentation in Brazilian clinical practice. The clinical picture has two layers. On the surface are the reactive crises of the familiopathic type, which follow conflicts at home within hours. Underneath are the slower swings of the thymopathic type, periods of days in which mood, energy and sleep change without a proportionate cause. The layers are not always easy to separate at interview, because a patient in an autonomous phase of irritability provokes more conflict and therefore has more reactive crises. A prospective record of mood 9 Research, Society and Development, v. 15, n. 10, e35151051799, 2026 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v15i10.51799 and sleep kept for some weeks, set beside a record of events at home, usually shows both patterns: sharp peaks tied to events, and a baseline that rises and falls independently of them. Three arguments support the proposal that this form predominates. The first is a matter of proportion: with heritability close to 46% and adversity reported by 71.4% of patients, overlap between the two is the expected case. The second is geneenvironment correlation, which has now been measured. The genetic load of BPD correlates with exposure to trauma and victimization, and assortative mating tends to concentrate liability and household instability in the same families (Skaug et al., 2022; Fan et al., 2025; Hall et al., 2026). A parent with an affective disorder passes on the vulnerability and, often enough, also provides an unstable home. The third argument concerns context: where childhood adversity is frequent, few children with affective vulnerability grow up in a protected environment. Once established, the mixed form tends to sustain itself. The instability of the patient increases tension in the household, a parent who is also affectively ill responds with less tolerance, and the conflicts that result feed the next crisis. Relatives of people with BPD have an increased risk of several psychiatric disorders (Hall et al., 2026), and parental psychopathology is among the early predictors of the disorder (Bozzatello et al., 2021), so the family that has to contain the patient is often a family that needs treatment. In this setting, treating the illness of a parent is itself a family intervention, and sometimes the one with the greatest effect on the patient. The mixed form is not homogeneous. Because the axes are continuous, the two loads are rarely equal, and it is useful to state which predominates. In a mixed form with familial predominance, the reactive layer accounts for most of the crises and the autonomous swings are mild, and removal from the home brings clear though incomplete relief. In a mixed form with thymic predominance, the swings dominate and the family contributes mainly by aggravating them. Stating the predominance does not create new types. It indicates which intervention is likely to produce the larger share of the improvement and which should be given priority when resources are limited. No Brazilian study has measured the proportion of the three types, and the claim of predominance is a clinical hypothesis. The contextual argument has some support in national data, since the notification system recorded 38,899 episodes of violence against children under ten in a single year, most of them at home (Malta et al., 2025). Figures of this kind show that exposure is frequent. They do not show that it is more frequent than elsewhere. The practical consequences are clear. In the mixed form, treating one axis alone yields a partial response. A stabilizer without work on the family leaves the trigger in place, and working on the family without a stabilizer leaves the autonomous oscillation untreated. The two interventions also act on different time scales. A change in the family, when it can be achieved, reduces reactive crises within weeks, whereas the effect proposed for lithium appears only after months. This difference is a source of error in follow-up: an early improvement may be credited to the drug when it is due to the change at home, and a late improvement may be credited to the family work when it is due to the drug. Keeping the two records described above throughout treatment is the simplest protection against this confusion. Partial response to a well-conducted treatment aimed at one axis should raise the suspicion of a mixed form. Whether the mixed form is more severe or runs a worse course than the pure types is an open question, and one that the cohort studies proposed in Section 3.9 could answer. 3.2.4 How the two loads interact The two axes are drawn at right angles in Figure 1, but they are not independent of each other, and the ways in which they interact are part of the model. Three can be distinguished. The first is transmission by the same parents. A parent with a mood disorder contributes genetic liability and, during 10 Research, Society and Development, v. 15, n. 10, e35151051799, 2026 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v15i10.51799 periods of illness, is less able to offer a stable and responsive home. The child receives both loads from one source. Assortative mating makes this more likely, because the partners of people with psychiatric disorders are themselves more often affected (Fan et al., 2025). The second is evocation. A child who is constitutionally more reactive is harder to soothe and more likely to draw impatient, punitive or dismissive responses from caregivers, and these responses increase the reactivity of the child. This circular process is what the biosocial model calls a transaction (Linehan, 1993), and it offers one explanation for the genetic correlation between borderline traits and reported trauma (Skaug et al., 2022). In families of this kind, it is difficult, and perhaps pointless, to ask which came first. The third is sensitivity. The genetic liability to BPD overlaps largely with the liability to post-traumatic stress disorder, with a genetic correlation of 0.90 in the Swedish family study (Hall et al., 2026). In the genome-wide study, post-traumatic stress disorder is likewise among the conditions with the strongest genetic correlation with BPD (Streit et al., 2026). Part of what is inherited, then, appears to be a heightened response to adverse events. If this is so, thymic load does more than add to family load: it influences how much damage a given exposure will do. Two siblings raised in the same harmful home may differ in outcome because they differ in sensitivity, and in a family study this difference is counted as non-shared environment. The apparent paradox of a disorder that is strongly associated with family adversity and yet shows almost no shared-environment effect may be explained in part by this mechanism. Taken together, these mechanisms explain why the mixed form should be the rule, and they clarify what a pure type means in the model. A pure familiopathic case is one in which exposure was severe enough to produce instability in a child of ordinary sensitivity. A pure thymopathic case is one in which sensitivity and autonomous dysregulation were sufficient even though the home was adequate. 3.2.5 Active and past family load The definition of the familiopathic type contains a condition that deserves emphasis: the family environment must still be acting on the patient. The literature on adversity is mostly retrospective and asks whether harm occurred in childhood (Porter et al., 2020). The model asks a different question, whether the harm is still occurring, because the answer changes what can be done. When the pathogenic environment is active, as with the adolescent who lives with an aggressor or the adult who remains bound to an invalidating family, each improvement obtained in treatment is undone at home, and the first task is to change that environment or to leave it. When the environment has ceased to act, what remains is the pattern of affect regulation that was learned in it. That pattern does not disappear by itself, but it is no longer being reinforced, and it is the proper object of structured psychotherapy, whose effect on symptom severity is established (Storebø et al., 2020). The same history of adversity can therefore correspond to two clinical situations with different priorities. This distinction also clarifies the criterion of removal from the environment. Improvement on removal is evidence of active load. Its absence in a patient with a documented history of adversity suggests either that the load is no longer active and the learned pattern has become autonomous, or that a thymic component is present. The two possibilities are separated by the remaining criteria, in particular the temporal pattern of affect and family history. 3.2.6 Provisional criteria Table 2 sets out the provisional criteria that distinguish the two pure types. 11 Research, Society and Development, v. 15, n. 10, e35151051799, 2026 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v15i10.51799 Table 2 - Provisional criteria for familiopathic and thymopathic types. Criterion Familiopathic Thymopathic 1. Family environment Documented and current adversity: maltreatment, neglect, domestic violence or chronic invalidation No relevant pathogenicity, or persistence of symptoms after removal from the home 2. Temporal pattern of affect Shifts lasting hours, tied to identifiable relational events Periods of days with changes in mood, energy and sleep without a proportionate trigger, superimposed on habitual reactivity 3. Family history and temperament No mood disorder in first-degree relatives; no autonomous mood episodes Bipolar disorder or recurrent depression in first-degree relatives; cyclothymic or irritable temperament from childhood or adolescence 4. Effect of removal from the environment Attenuation of symptoms, with relapses on return Little effect Source: Authors (2026). The criteria in Table 2 were derived from the logic of the model and have not been validated. They are offered as a starting point for operationalization. A patient who meets criteria in both columns is classified as mixed. 3.2.7 Illustrative prototypes Three schematic prototypes, constructed for illustration and not drawn from individual patients, show how the criteria combine. The familiopathic prototype is an adolescent whose outbursts and self-injury began after a change in the composition of the household. The episodes follow conflicts at home within hours, and there is no mood disorder among close relatives. During a two-week admission the symptoms subside without any change in medication, and they return in the first days after discharge. The thymopathic prototype is a young adult from a stable and attentive family, with a parent treated for recurrent depression. Since early adolescence there have been stretches of several days of irritability, reduced sleep and increased energy that no event explains, in addition to the usual sensitivity to rejection. Admissions and changes of residence make little difference. The mixed prototype combines the two histories: a parent with an untreated mood disorder, a household marked by conflict and unpredictability, and a patient whose reactive crises are superimposed on autonomous swings. Removal from the home brings partial relief, and so does a mood stabilizer, but neither is sufficient alone. 3.2.8 Relation to neighboring diagnoses Each type borders on a different diagnosis, and the genetic data helps to draw these borders. The thymopathic type borders on the mood disorders. Its closest genetic relative is depressive disorder, and its relation to bipolar disorder is one of partial overlap, as discussed above. The familiopathic type borders on the trauma-related disorders. Post-traumatic stress disorder showed the most pronounced familial co-aggregation with BPD across all types of relatives, a genetic correlation of 0.90, and the largest contribution of shared environment among the psychiatric conditions studied (Hall et al., 2026). The two diagnoses share symptoms, and part of the overlap may reflect the difficulty of telling them apart. For the present model, the point is that a patient whose instability is driven by ongoing maltreatment stands where the two diagnoses meet, and the question of which label applies matters less than the recognition that the harmful exposure has not ended. A third border is not covered by either axis. BPD also shares genetic liability with attention-deficit/hyperactivity disorder and with antisocial behavior (Streit et al., 2026), and attention-deficit/hyperactivity disorder appears among the early factors common to BPD and bipolar disorder (Durdurak et al., 2022). This suggests a route to the disorder through constitutional 12 Research, Society and Development, v. 15, n. 10, e35151051799, 2026 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v15i10.51799 impulsivity that is neither affective nor familial. The model does not address it, and patients in whom impulsivity has dominated the picture since childhood may fit poorly into any of the three types. 3.3 The non-parent maternal partner Among the family factors, the model gives the greatest weight to one specific situation: a partner of the mother, with no biological tie to the child, comes to live in the household and begins to maltreat the child. The literature on child maltreatment confirms that this situation carries a high risk of severe harm. In a case-control study of deaths from inflicted injury in young children, living with an unrelated adult, in most cases the mother's boyfriend, raised the risk about 50-fold in comparison with households with two biological parents. In the same study, children living with a single parent and no other adult were at no greater risk, which shows that the danger lies in the presence of the unrelated adult and not in the absence of a parent (Schnitzer & Ewigman, 2005). More recently, an analysis of 2,414 child abuse-related homicides in the United States found that deaths precipitated by harsh physical punishment had 2.4 times the odds of involving the mother's male companion rather than the father. Deaths of this kind also more often involved a child with a previous nonfatal injury and another adult who took part in the incident or knew of earlier abuse (Wilson et al., 2023). This configuration also shows how family adversity can be specific to one child. A partner who is hostile to a stepchild may treat his own children in the same household quite differently. For the stepchild the home is pathogenic, for the half-siblings it may not be, and a family study will register the difference as environment that is not shared. The situation that the model regards as the most destructive is, for this reason, one that register studies are least able to detect as an effect of the family. Two cautions apply. The first is that the risk is relative. The Brazilian notification system (SINAN) recorded 38,899 notifications of violence against children aged 0 to 9 years in 2022, and 88.3% of the episodes took place at home. The aggressor was the mother in 51.7% of cases, the father in 40% and the stepfather in 6.2%, with more than one aggressor possible in the same notification (Malta et al., 2025). Most of the violence is committed by biological parents, and most unrelated adults who live with children do them no harm. The second caution is that the studies cited measure injury and death. The link between this configuration and later BPD is an inference that still must be tested. The psychopathological hypothesis is that this situation adds three injuries. The child is maltreated directly, loses the mother as a protective figure, and finds that the complaint is not believed. The last two are forms of invalidation at their most severe, and the finding that another adult often knew of earlier abuse gives empirical weight to the idea of failed protection (Wilson et al., 2023). The formulation must rest on that failure and not on the mother's choice of partner, since she is frequently a victim of violence in the same household. 3.4 Therapeutic implications The type indicates where the decisive intervention lies. Table 3 compares the three types on the dimensions that guide treatment. Table 3 - Comparison of the three types and their therapeutic consequences. Dimension Familiopathic Thymopathic Mixed Source of instability Active family environment Constitutional affective dysregulation Both Pattern of affect Reactive, hours Autonomous, days Reactive on an autonomous background Mood disorder in the family Absent Present Present Little effect Partial improvement Effect of removal from the home Improvement 13 Research, Society and Development, v. 15, n. 10, e35151051799, 2026 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v15i10.51799 Dimension Familiopathic Thymopathic Mixed Decisive intervention Family and protection Mood stabilizer Both, in parallel Antipsychotic Symptomatic, while exposure lasts Bridge until lithium takes effect Symptomatic and bridge Lithium Not indicated by the model Response expected at about 6 months Indicated for the thymic component Source: Authors (2026). Table 3 must be read against what is known about treatment in BPD. Structured psychotherapy is the first-line treatment. In the Cochrane review of 75 randomized trials, psychological therapies reduced the severity of symptoms and suicidality in comparison with usual treatment, and no modality proved superior to the others (Storebø et al., 2020). The guideline of the American Psychiatric Association recommends a structured psychotherapy for every patient and suggests that any medication be adjunctive, limited in time and directed at a measurable target symptom (Keepers et al., 2024). The syntheses of drug trials point the same way. The Cochrane review of pharmacological interventions, with 46 trials and 2,769 participants, concluded that no drug treats the core psychopathology of the disorder (Stoffers-Winterling et al., 2022). An independent meta-analysis reached a similar conclusion (Gartlehner et al., 2021), and a secondary analysis of the Cochrane data examined the depressive and psychotic symptoms that accompany BPD (Pereira Ribeiro et al., 2025). A network metaanalysis of 35 trials found that topiramate, lamotrigine and aripiprazole reduced hostility and anger, while haloperidol was among the drugs supported only by low-certainty evidence and was not recommended as a priority in unselected samples (Gerolymos et al., 2026). The model is consistent with this literature in one respect and departs from it in another. It is consistent in treating the antipsychotic as a symptomatic and temporary resource, which is what the guideline recommends for any medication in BPD. It departs from the literature in proposing a long-term mood stabilizer for a subgroup defined by affective markers. That proposal does not contradict the trials, which enrolled unselected samples, but it is not supported by them either. The typology also offers a reading of two findings that are otherwise hard to interpret. The first is that no psychotherapy has proved superior to the others (Storebø et al., 2020). If patients differ in the source of their instability, treatments that differ in technique but are applied to the same mixture of patients will produce similar average results, and differences would emerge only when patients are matched to interventions. The second is the pattern of drug effects. The agents with some support, topiramate, lamotrigine and aripiprazole, act on hostility and anger and not on the core of the disorder (Stoffers-Winterling et al., 2022; Gerolymos et al., 2026). Two of the three are anticonvulsants, a class that includes established mood stabilizers, which is at least compatible with the idea that a share of patients has an affective dysregulation responsive to this class, although the effect seen in unselected samples is small and was measured over weeks. The emphasis of psychotherapy also varies with the type, although none of the three dispenses with it. In the familiopathic type, psychoeducation, which the guideline recommends for every patient (Keepers et al., 2024), is extended to the relatives, and their participation becomes a condition of treatment. In the thymopathic type, psychotherapy addresses the consequences of long-standing affective instability for identity and relationships while the stabilizer is given time to act. Suicidal behavior cuts across the three types and needs attention in all of them. Self-harm shares a large part of its genetic basis with BPD, with a genetic correlation of 0.80, yet the association of BPD with self-harm and suicide was much stronger within individuals than between relatives, which indicates that circumstances specific to the person weigh heavily on these outcomes (Hall et al., 2026). The practical reading is that the risk cannot be deduced from the type. Psychotherapy is the intervention with a demonstrated effect on suicidality (Storebø et al., 2020), and the suggestion that lithium may reduce suicidal symptoms in BPD rests on clinician ratings in a single small trial (Links et al., 1990). Three further points concern safety. The choice of antipsychotic needs to be justified. The clinical formulation from 14 Research, Society and Development, v. 15, n. 10, e35151051799, 2026 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v15i10.51799 which the model derives refers to haloperidol at full dose, yet the available evidence places haloperidol below aripiprazole in certainty, and any protocol must state dose and duration and provide for monitoring of extrapyramidal symptoms, tardive dyskinesia and the QT interval. Lithium requires particular care in this population. Its narrow therapeutic index is a hazard in impulsive patients at risk of self-poisoning, which calls for dispensing in small quantities, regular serum levels and monitoring of renal and thyroid function, and its teratogenic risk has to be weighed in women of childbearing age. Finally, in the familiopathic and mixed types, when the patient is a child or an adolescent, the assessment of risk and the activation of the protection network come before any adjustment of dose. 3.5 Classification algorithm The classification can be made in five steps, set out in Table 4. It is not completed in a single interview. The first assessment yields a provisional type, which is reviewed as information accumulates from the patient, from relatives interviewed separately, from previous records and, when there is an admission, from direct observation. Table 4 - Five-step algorithm for classifying a patient with borderline personality disorder. Step Action What the finding indicates 1 Confirm the diagnosis of BPD by operational criteria and exclude a frank manic or hypomanic episode A frank episode defines comorbid bipolar disorder 2 Map current family exposure: who lives in the household, presence of violence, neglect or chronic invalidation, and whether the patient still lives there Familiopathic axis 3 Search for autonomous affective oscillation: periods of days with changes in mood, energy and sleep without a proportionate trigger Thymopathic axis 4 Record mood disorder in first-degree relatives and the previous temperament of the patient Thymopathic axis 5 Observe the effect of removal from the environment, when it occurs, through hospitalization or change of residence Familiopathic axis Source: Authors (2026). Positive findings in steps 2 and 5 of Table 4, with negative findings in steps 3 and 4, indicate the familiopathic type. The reverse pattern indicates the thymopathic type, and markers on both axes indicate the mixed form. When the information is insufficient or contradictory, the type is recorded as undetermined and the assessment is repeated after about three months of follow-up. Each step has its own sources of information and its own pitfalls, which are summarized in Table 5. Table 5 - Sources of information and main pitfall at each step of the algorithm. Step Sources of information Main pitfall 1 Clinical interview with operational criteria; longitudinal history of mood episodes Taking brief reactive shifts for hypomania, or overlooking a frank episode 2 Interview with the patient alone and with relatives separately; timeline of changes in household composition; school, socialservice and medical records Asking only about physical and sexual violence and missing emotional abuse, neglect and invalidation; treating past adversity as current 3 Prospective record of mood, energy and sleep for some weeks, set beside a record of events Relying on retrospective report, in which every change of mood is attributed to an event 4 Family history in first-degree relatives, including treatment received; account of temperament from someone who knew the patient as a child Reading a positive family history as purely genetic, when an ill parent also shapes the home 5 Course of symptoms during admission or change of residence and after the return Attributing to removal an improvement due to the routine of the ward, supervised medication or the attention of staff Source: Authors (2026). 15 Research, Society and Development, v. 15, n. 10, e35151051799, 2026 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v15i10.51799 As Table 5 indicates, the first step guards the borders of the typology. The distinction that Paris and Black (2015) drew between brief reactive shifts and sustained changes of mood is applied here to separate BPD from bipolar disorder, and a frank manic or hypomanic episode takes the patient out of the scheme, since the treatment of bipolar disorder then takes precedence. The borders with post-traumatic stress disorder and with attention-deficit/hyperactivity disorder, discussed in Section 3.2.8, are examined at the same point. The second step requires more than a question about violence at home. The adversities most strongly associated with BPD are emotional abuse and neglect (Porter et al., 2020), and invalidation without overt abuse carries its own effect (Lee et al., 2022), so the interview has to cover how the family responds to the emotions of the patient and not only whether anyone has been hit. The patient is seen alone for this part, because an aggressor may be the accompanying relative. A timeline of who lived in the household at each period is drawn, with attention to the arrival of new adults. The interviewer also establishes whether the exposure is current, for the reasons given in Section 3.2.5. The third and fourth steps assess thymic load, and they check each other. Autonomous oscillation is best documented prospectively, because in retrospect patients and relatives tend to find an event to explain every change of mood. Sleep is the most reliable item in the record, for the reasons given in Section 3.2.2. Family history is informative but ambiguous. A parent with a mood disorder indicates genetic liability, yet the same parent may also have provided an unstable home, so a positive family history together with a difficult family environment point to the mixed form and not to the thymopathic type alone. The fifth step uses separation from the family as a natural experiment, and it must be interpreted with care. An admission removes the patient from home, but it also provides routine, supervised medication and the attention of staff, any of which may reduce symptoms. Improvement during admission is therefore weak evidence by itself. The observation gains weight when symptoms return soon after discharge with no change in treatment, and more still when the sequence repeats across admissions. Removal is indicated by clinical need or by risk and is never arranged as a diagnostic test. Two errors of classification are possible, and their costs differ. Classifying a mixed case as familiopathic leaves the thymic component untreated, which delays an intervention that is itself still hypothetical. Classifying a mixed or familiopathic case as thymopathic leaves an active harmful environment unrecognized, and in a child or adolescent this may mean continued maltreatment. The second error is the more serious. For this reason, the investigation of family exposure comes before the assessment of thymic load in Table 4, and when doubt remains the working assumption is the mixed form, which keeps both lines of intervention open. The classification is revised in the light of the response to treatment. If reactive crises persist after the family situation has demonstrably changed, thymic load is reassessed. If instability persists after an adequate period on a mood stabilizer, family load is reassessed, with particular attention to exposures that were denied at the first interview. A type is a working hypothesis about the patient, and a response that contradicts it is information. The algorithm has not been validated. Its first empirical test is agreement between independent raters, which must be shown before any of the hypotheses in Section 3.9 can be examined. 3.6 Early identification and prevention Because both loads are present long before the diagnosis is made, the typology has implications for early identification. The factors that precede BPD in prospective studies include family history of psychopathology, affective instability, sleep disturbance and childhood adversity (Durdurak et al., 2022), and reviews of early detection place parental psychopathology, maltreatment and the temperament of the child among the main predictors (Bozzatello et al., 2021). These are the same elements that define the two axes. An adolescent with emerging borderline features can therefore be assessed for family load and thymic 16 Research, Society and Development, v. 15, n. 10, e35151051799, 2026 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v15i10.51799 load before the full disorder is established. The preventive consequences differ by axis. Where family load predominates, prevention is a matter of child protection. The Brazilian notification data show that most violence against young children takes place at home and that its form changes with age, with neglect predominating in the first year of life and sexual violence appearing between two and five years (Malta et al., 2025). Services that receive these notifications can act on the family factor years before a personality diagnosis is considered. Where thymic load predominates, prevention consists of follow-up of the children of parents with mood disorders, with attention to early changes in sleep and affect. In both situations, involving relatives is part of the intervention, and programs for family members already exist that can be adapted to this purpose (Guillén et al., 2021). 3.7 Evidence map Table 6 grades each thesis of the model according to the support found in the literature. Table 6 - Theses of the model and level of empirical support. Thesis Support Main sources BPD has a heritable constitutional component Strong Skoglund et al. (2021); Streit et al. (2026) Family adversity is associated with BPD Strong for association, moderate for causality Porter et al. (2020); Lee et al. (2022); Skaug et al. (2022) The pure familial form is rare Indirect, consistent with minimal shared environment Hall et al. (2026); Skaug et al. (2022) A subtype with an affective basis exists Moderate: high genetic correlation with depression, 0.35 with bipolar disorder Streit et al. (2026); Hall et al. (2026); Durdurak et al. (2022) Thymic load modulates the effect of family load Indirect: genetic overlap with posttraumatic stress disorder and with reported trauma Hall et al. (2026); Streit et al. (2026); Skaug et al. (2022) The mixed form predominates in Brazil No data Clinical hypothesis The non-parent maternal partner is the most severe family factor Strong for injury and death, absent for BPD Schnitzer & Ewigman (2005); Wilson et al. (2023) The antipsychotic is only symptomatic Consistent with guidelines Keepers et al. (2024); Gerolymos et al. (2026) Lithium is effective in the thymopathic subtype after 6 months Weak: one small crossover trial Links et al. (1990); Almeida & Sanches (2024); Gallicchio et al. (2026) Family intervention changes the course in the familiopathic type Weak to moderate Guillén et al. (2021) Source: Authors (2026). Table 6 shows an asymmetry. The etiological theses rest on strong or moderate evidence, drawn from large genetic and epidemiological studies. The therapeutic theses rest on weak evidence or on none. The model is therefore better established as a way of understanding patients than as a guide to prescribing, and the research agenda that follows from it is mainly therapeutic. 3.8 Limitations and objections The typology arises from clinical observation and has not been validated, and several objections can be raised against it. The most serious concerns the familiopathic type. Register studies do not detect an appreciable effect of the home shared by siblings (Skoglund et al., 2021; Hall et al., 2026), so the type can be sustained only if adversity is specific to one child. In addition, if the link between trauma and borderline traits is largely genetic (Skaug et al., 2022), many cases that seem familiopathic may be mixed forms in which the thymic component has gone unrecognized. The measurement of adversity is itself uncertain, since the association falls from an odds ratio of 13.91 to about 2.6 outside case-control designs (Porter et al., 2020). 17 Research, Society and Development, v. 15, n. 10, e35151051799, 2026 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v15i10.51799 The thymopathic type raises a different concern. Paris and Black (2015) warned that treating reactive instability as if it were bipolar disorder deprives patients of the psychotherapy that helps them. The type needs strict criteria, or it will become a justification for polypharmacy. The therapeutic hypothesis attached to it depends on a single small trial and must be presented to patients as such. A third objection is that the two axes do not exhaust the routes to the disorder. The genetic overlap of BPD with attention-deficit/hyperactivity disorder and antisocial behavior (Streit et al., 2026) points to a constitutional impulsivity that the model leaves out, and a complete typology may need a third axis. There are also limits to generalization. The genome-wide study was restricted to people of European ancestry and the register studies to Sweden, where diagnoses are made in specialist care and probably reflect more severe cases (Streit et al., 2026; Hall et al., 2026). The symptom-based subtypes described so far lack longitudinal validation (Triantafyllou et al., 2025), which means there is no established classification against which the present one can be compared. No Brazilian data on the proposed types exist. Lastly, the review that supports the essay was directed and not systematic, and relevant studies may have been missed. 3.9 Testable hypotheses The model yields four predictions, each of which could show it to be wrong. H1. In patients with BPD, a family history of mood disorder and autonomous affective oscillation predict response to lithium at 6 to 12 months, and the absence of these markers predicts non-response. H2. In the familiopathic type, symptom severity follows exposure to the family environment and falls during removal, more than in the thymopathic type. H3. In Brazilian clinical samples, the mixed form exceeds 50% of cases. H4. Maltreatment by a non-parent maternal partner is associated with greater severity than other adversities, controlling for type of abuse. A trial designed to test the first hypothesis would classify patients at baseline with the algorithm in Table 4, randomize them within each type to lithium or placebo added to structured psychotherapy, and follow them for at least twelve months, with affective instability, anger and suicidal behavior as outcomes. The prediction is an interaction between type and treatment: benefit in the thymopathic and mixed groups and none in the familiopathic group. A result without this interaction would count against the model. The second hypothesis can be examined wherever patients are admitted, since each admission is a natural experiment. Symptom ratings at admission, at discharge and some weeks after the return home, compared between types, would show whether severity follows exposure to the family as predicted. The third and fourth hypotheses require a clinical cohort assessed with structured interviews for childhood adversity and family history. For the fourth, the composition of the household at each age must be recorded, information that standard questionnaires on adversity do not collect. Polygenic scores, which already explain a measurable share of liability (Streit et al., 2026), could later be added as an independent marker of thymic load. This would allow the hypothesis of sensitivity described in Section 3.2.4 to be tested directly, by asking whether the effect of a given adversity on symptoms increases with the genetic score. 4. Final Considerations The distinction between familiopathic, thymopathic and mixed borderline gives the clinician a simple question to begin with: does the instability of this patient come from the home, from mood, or from both? The answer indicates where treatment 18 Research, Society and Development, v. 15, n. 10, e35151051799, 2026 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v15i10.51799 should concentrate and explains why medication alone fails when the family continues to make the patient ill. Recent literature gives the model two solid supports and one constraint. The supports are the polygenic basis of the disorder, with an inherited liability that is largely affective, and the consistent association with adversity and invalidation. The constraint is the small contribution of shared environment, which explains why the pure familial form is rare and why the mixed form is the one to expect. Two refinements make the model more precise. The axes interact, since part of the inherited liability seems to be a greater sensitivity to adversity, and this helps to reconcile the weight of family adversity with the absence of shared-environment effects. The family load also has a time dimension: an environment that is still acting calls for protection and change, whereas an environment that has ceased to act leaves a learned pattern that is the object of psychotherapy. The pharmacological part of the model goes beyond current evidence and is presented as a hypothesis. Future studies should operationalize the criteria, test the four predictions in Brazilian samples and determine, in trials stratified by type, whether lithium benefits the thymopathic subgroup. References Akiskal, H. S. (2004). Demystifying borderline personality: Critique of the concept and unorthodox reflections on its natural kinship with the bipolar spectrum. Acta Psychiatrica Scandinavica, 110(6), 401-407. Almeida, T. M., & Sanches, M. (2024). Rebuilding the evidence on the use of lithium for borderline personality disorder. Clinical Neuropsychiatry, 21(6). https://doi.org/10.36131/cnfioritieditore20240608 Bozzatello, P., Garbarini, C., Rocca, P., & Bellino, S. (2021). Borderline personality disorder: Risk factors and early detection. Diagnostics, 11(11), 2142. https://doi.org/10.3390/diagnostics11112142 Durdurak, B. B., Altaweel, N., Upthegrove, R., & Marwaha, S. (2022). Understanding the development of bipolar disorder and borderline personality disorder in young people: A meta-review of systematic reviews. Psychological Medicine, 52(16), 3769-3782. Fan, C. C., Dehkordi, S. R., Border, R., Shao, L., Xu, B., Loughnan, R., et al. (2025). Spousal correlations for nine psychiatric disorders are consistent across cultures and persistent over generations. Nature Human Behaviour, 9, 2539-2547. Gallicchio, V. S., Simon, L., Jafarmadar, A., Ramanathan, P., Owens-Charles, C. R., Singer, S., & Warlick, H., IV. (2026). Lithium in borderline personality disorder: Neuroplasticity, anti-inflammatory action, and therapeutic potential. Frontiers in Behavioral Neuroscience, 20, 1707007. https://doi.org/10.3389/fnbeh.2026.1707007 Gartlehner, G., Crotty, K., Kennedy, S., Edlund, M. J., Ali, R., Siddiqui, M., et al. (2021). Pharmacological treatments for borderline personality disorder: A systematic review and meta-analysis. CNS Drugs, 35(10), 1053-1067. https://doi.org/10.1007/s40263-021-00855-4 Gerolymos, C., Garosi, A., Boyer, L., Yon, D. K., Rahmati, M., Gavaudan, M., et al. (2026). Efficacy and safety of pharmacological treatments in borderline personality disorder: A systematic review and network meta-analysis. Molecular Psychiatry, 31(6), 3216-3228. https://doi.org/10.1038/s41380-026-03451-4 Guillén, V., Díaz-García, A., Mira, A., García-Palacios, A., Escrivá-Martínez, T., Baños, R., & Botella, C. (2021). Interventions for family members and carers of patients with borderline personality disorder: A systematic review. Family Process, 60(1), 134-144. https://doi.org/10.1111/famp.12537 Hall, A. S. M., Musliner, K. L., Debost, J.-C. P., Vilhjálmsson, B. J., Chang, Z., D'Onofrio, B. M., Lichtenstein, P., Kuja-Halkola, R., et al. (2026). Borderline personality disorder and other psychiatric, somatic, and behavioral conditions: A nationwide family study. Translational Psychiatry, 16, 257. https://doi.org/10.1038/s41398-026-04001-w Keepers, G. A., Fochtmann, L. J., Anzia, J. M., Benjamin, S., Lyness, J. M., Mojtabai, R., Servis, M., Choi-Kain, L., Nelson, K. J., Oldham, J. M., Sharp, C., Degenhardt, A., Hong, S. H., & Medicus, J. (2024). The American Psychiatric Association practice guideline for the treatment of patients with borderline personality disorder. American Journal of Psychiatry, 181(11), 1024-1028. Kleindienst, N., Vonderlin, R., Bohus, M., & Lis, S. (2021). Childhood adversity and borderline personality disorder: Analyses complementing the meta-analysis by Porter et al. (2020). Acta Psychiatrica Scandinavica, 143(2), 183-184. https://doi.org/10.1111/acps.13256 Lee, S. S. M., Keng, S.-L., Yeo, G. C., & Hong, R. Y. (2022). Parental invalidation and its associations with borderline personality disorder symptoms: A multivariate meta-analysis. Personality Disorders: Theory, Research, and Treatment, 13(6), 572-582. https://doi.org/10.1037/per0000523 Leichsenring, F., Fonagy, P., Heim, N., Kernberg, O. F., Leweke, F., Luyten, P., et al. (2024). Borderline personality disorder: A comprehensive review of diagnosis and clinical presentation, etiology, treatment, and current controversies. World Psychiatry, 23(1), 4-25. https://doi.org/10.1002/wps.21156 Linehan, M. M. (1993). Cognitive-behavioral treatment of borderline personality disorder. Guilford Press. Links, P. S., Steiner, M., Boiago, I., & Irwin, D. (1990). Lithium therapy for borderline patients: Preliminary findings. Journal of Personality Disorders, 4(2), 173-181. https://doi.org/10.1521/pedi.1990.4.2.173 19 Research, Society and Development, v. 15, n. 10, e35151051799, 2026 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v15i10.51799 Malta, D. C., et al. (2025). Fatores associados à notificação de violência na infância no Brasil. Ciência & Saúde Coletiva, 30(2), e00572024. https://doi.org/10.1590/1413-81232025302.00572024 Ogassawara, D. et al. (2025). Trilhas metodológicas para a revisão narrativa: orientações pragmáticas para sua elaboração. Ensino & Pesquisa, 23(3). https://doi.org/10.33871/23594381.2025.23.3.10317 Paris, J., & Black, D. W. (2015). Borderline personality disorder and bipolar disorder: What is the difference and why does it matter? Journal of Nervous and Mental Disease, 203(1), 3-7. https://doi.org/10.1097/NMD.0000000000000225 Pereira, A. S. et al. (2018). Metodologia da pesquisa científica. [Free ebook]. Santa Maria. Editora da UFSM. https://orcid.org/0000-0002-9813-0457 Pereira Ribeiro, J., Juul, S., Kongerslev, M. T., Jørgensen, M. S., Völlm, B. A., Edemann-Callesen, H., et al. (2025). Pharmacological interventions for cooccurring psychopathology in people with borderline personality disorder: Secondary analysis of the Cochrane systematic review with meta-analyses. British Journal of Psychiatry, 226, 226-237. https://doi.org/10.1192/bjp.2024.172 Porter, C., Palmier-Claus, J., Branitsky, A., Mansell, W., Warwick, H., & Varese, F. (2020). Childhood adversity and borderline personality disorder: A metaanalysis. Acta Psychiatrica Scandinavica, 141(1), 6-20. https://doi.org/10.1111/acps.13118 Ramos-Suárez, I., Guerrero-Jiménez, M., Cervilla, J. A., & Gutiérrez, B. (2026). Epidemiology of borderline personality disorder in the general population: Prevalence, sociodemographic factors, and comorbidities. A systematic review and meta-analysis. Journal of Affective Disorders, 394, 120482. https://doi.org/10.1016/j.jad.2025.120482 Risemberg, R. I. C. et al. (2026a). A importância da metodologia científica no desenvolvimento de artigos científicos. E-Acadêmica, 7(1), e0171675. https://eacademica.org/eacademica/article/view/675. Risemberg, R. I. C., Wakin, M., & Shitsuka, R. (2026b). Qualitativa, quantitativa ou quali-quantitativa: Como identificar de modo rápido a abordagem de um artigo científico? E-Acadêmica, 7(2), e0172702. https://doi.org/10.52076/eacad-v7i2.702 Rother, E. T. (2007). Revisão sistemática https://scielo.br/j/ape/a/z7zZ4Z4GwYV6FR7S9FHTByr/ X revisão narrativa. Acta Paulista de Enfermagem, 20(2), v-vi. Saccaro, L. F., Schilliger, Z., Dayer, A., Perroud, N., & Piguet, C. (2021). Inflammation, anxiety, and stress in bipolar disorder and borderline personality disorder: A narrative review. Neuroscience & Biobehavioral Reviews, 127, 184-192. https://doi.org/10.1016/j.neubiorev.2021.04.017 Santos de Souza, L. (2025). Ensaios teórico-metodológicos: ensino e pesquisas em educação, cultura, análise cognitiva e difusão. Editora CRV. Schnitzer, P. G., & Ewigman, B. G. (2005). Child deaths resulting from inflicted injuries: Household risk factors and perpetrator characteristics. Pediatrics, 116(5), e687-e693. https://doi.org/10.1542/peds.2005-0296 Skaug, E., Czajkowski, N. O., Waaktaar, T., & Torgersen, S. (2022). Childhood trauma and borderline personality disorder traits: A discordant twin study. Journal of Psychopathology and Clinical Science, 131, 365-374. Skoglund, C., Tiger, A., Rück, C., Petrovic, P., Asherson, P., Hellner, C., Mataix-Cols, D., & Kuja-Halkola, R. (2021). Familial risk and heritability of diagnosed borderline personality disorder: A register study of the Swedish population. Molecular Psychiatry, 26(3), 999-1008. https://doi.org/10.1038/s41380-019-04420 Stoffers-Winterling, J. M., Storebø, O. J., Pereira Ribeiro, J., Kongerslev, M. T., Völlm, B. A., Mattivi, J. T., Faltinsen, E., Todorovac, A., Jørgensen, M. S., Callesen, H. E., Sales, C. P., Schaug, J. P., Simonsen, E., & Lieb, K. (2022). Pharmacological interventions for people with borderline personality disorder. Cochrane Database of Systematic Reviews, 2022(11), CD012956. https://doi.org/10.1002/14651858.CD012956.pub2 Storebø, O. J., Stoffers-Winterling, J. M., Völlm, B. A., Kongerslev, M. T., Mattivi, J. T., Jørgensen, M. S., Faltinsen, E., Todorovac, A., Sales, C. P., Callesen, H. E., Lieb, K., & Simonsen, E. (2020). Psychological therapies for people with borderline personality disorder. Cochrane Database of Systematic Reviews, 2020(5), CD012955. https://doi.org/10.1002/14651858.CD012955.pub2 Streit, F., Awasthi, S., Hall, A. S. M., Braun, A., Niarchou, M., Marouli, E., et al. (2026). Genome-wide association analyses of borderline personality disorder identify 11 loci and highlight shared risk with mental and somatic disorders. Nature Genetics. https://doi.org/10.1038/s41588-026-02654-3 Triantafyllou, A., Stefanatou, P., Konstantakopoulos, G., et al. (2025). Unveiling the layers of borderline personality disorder: A systematic review of clinical subtypes. Behavioral Sciences, 15(7), 928. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12292403/ Wilson, R. F., Afifi, T. O., et al. (2023). Child abuse-related homicides precipitated by caregiver use of harsh physical punishment. Child Abuse & Neglect, 135, 105953. https://doi.org/10.1016/j.chiabu.2022.105953 20