Diagnoses of double heterozygous NF1 variants and dual RASopathy Authors: Daniela Angelova-Toshkina, Denny Schanze, Pia Vaassen, Michael C Frühwald, Thorsten Rosenbaum, Martin Zenker, Michaela Kuhlen Published: 2026 ABSTRACT AbstractGenetic testing in neurofibromatosis type 1 (NF1) occasionally reveals two heterozygous NF1 variants in the same individual. Correct interpretation hinges on allelic phasing, elucidation of somatic second hits in lesions, and distinction from bona fide dual RASopathy diagnoses. We provide a focused, critical review of the primary literature on (i) cis doublets in NF1, (ii) purported germline trans configurations—particularly in spinal neurofibromatosis, (iii) biallelic somatic inactivation of NF1 in café-au-lait macules, tumors, and juvenile myelomonocytic leukemia, and (iv) confirmed cases of dual RASopathy. We also include an index patient with two heterozygous NF1 variants to illustrate clinical challenges. Phase-proven cis doublets are rare, often closely spaced, and generally accompany classic NF1 without reproducible phenotype escalation. Claims of germline trans NF1 pathogenic variants weaken on reappraisal: second alleles reported in spinal neurofibromatosis typically do not meet current pathogenicity standards, while Nf1−/− mouse embryonic lethality underscores the biological implausibility of constitutive biallelic loss. In contrast, biallelic somatic NF1 inactivation is pervasive across NF1 lesions, including café-au-lait macules, plexiform and cutaneous neurofibromas, and juvenile myelomonocytic leukemia, due to mitotic recombination (copy-neutral loss-of-heterozygosity), microdeletions, or a second variant. Confirmed dual RASopathy diagnoses (most often NF1 plus PTPN11) are rare and produce blended systemic phenotypes. Interpretation of double NF1 findings should prioritize independent American College of Medical Genetics and Genomics/Association for Molecular Pathology classification of each variant, rigorous phasing (segregation/RNA/allele-specific methods), and tumor-focused analysis where applicable. Our index case and synthesis support a practical workflow that clarifies counseling and helps avoid over-calling dual pathogenicity. Introduction Neurofibromatosis type 1 (NF1; MIM #162200, ORPHA:636) is a common RASopathy following an autosomal dominant inheritance pattern, near-complete penetrance, and striking variability of expression. Approximately half of the cases arise from de novo variants, and inter-individual courses span pigment-predominant phenotypes to heavy internal tumor burdens, with additional morbidity due to neurocognitive and behavioral difficulties. Despite decades of study, genotype–phenotype correlations remain limited at the individual level. Aside from a few established associations [e.g., 17q11.2 microdeletions and the recurrent alleles NM_000267.3(NF1):c.5425 C > T (p.Arg1809Cys) and NM_001042492.3(NF1):c.2970_2972del (p.Met992del)] the identity or class of the germline NF1variant does not reliably predict clinical outcomes such as the occurrence or burden of tumors, age at onset of optic pathway glioma, neurocognitive/behavioral difficulties, or vascular complications. This uncertainty complicates counseling and surveillance planning for families [1–3]. Molecularly, NF1 is a formidable diagnostic challenge. More than 3,000 distinct disease-causing germline variants have been described in the literature, predominantly single-nucleotide substitutions and small indels, with a significant minority of microdeletions and complex splice-altering changes [1]. Large public resources curate many thousands of additional sequence records. ClinVar lists > 16,000 variants mapping to NF1 [single-nucleotide variants/indels and overlapping copy number variations (CNVs); access date November 2025], while a locus-specific NF1 database (www.LOVD.nl/NF1) provides transcript-resolved entries (e.g. NM_000267.3; NM_001042492.3). The possibility of mosaicism further muddles interpretation. In ~ 5–10% of individuals, NF1 results from recurrent 17q11.2 microdeletions that remove the entire NF1 gene plus flanking genes [4–6]. The most frequent, type-1 deletions (~ 1.4 Mb), arise by non-allelic homologous recombination between NF1 and flanking repetitive sequences (termed REPs) and typically remove ~ 14 protein-coding genes including SUZ12. They are usually constitutional (germline) and account for ~ 70–80% of NF1 microdeletions [7]. Type-2 deletions (~ 1.2 Mb) result from recombination between SUZ12 and SUZ12P1and are often post-zygotic (mosaic) [8]. Type-3 (~ 1.0 Mb) and atypical deletions are less common with variable breakpoints [9]. On average, microdeletion genotypes associate with earlier and heavier tumor burden and additional dysmorphic/neurodevelopmental features. Awareness of these classes informs testing [e.g., multiplex ligation-dependent probe amplification (MLPA)/next-generation sequencing (NGS) CNV calling]. In response, comprehensive testing strategies that pair high coverage sequencing with copy-number analyses and, when indicated, dosage analysis and RNA studies for cryptic splice effects have become routine in many centers [10]. The 2021 international revision of diagnostic criteria explicitly integrated molecular testing and clarified the boundary with Legius syndrome (SPRED1-related; MIM #611431, ORPHA:137605), which is a key differential diagnosis in pigment-only presentations [11]. Contemporary multigene testing has highlighted that some individuals carry pathogenic or likely pathogenic (subsequently referred to as pathogenic) variants in more than one cancer-predisposition gene (variously termed multilocus inherited neoplasia or double heterozygosity). Although uncommon, such multilocus genotypes recur across cohorts and often influence surveillance and counseling—underscoring the need for careful variant interpretation in context [12, 13]. Against this backdrop, encountering two NF1 variants in one individual raises the central question of allelic phase: do the variants reside on the same allele (cis; monoallelic) or on opposite alleles (trans; biallelic)? Establishing the phase is a foundational informational step and typically relies on segregation studies (parents), allele-specific polymerase chain reaction (PCR)/cloning, RNA/complementary DNA analysis for splice effects, or long-read/haplotype-based approaches when needed. The clinical importance of allelic status within a single gene is well illustrated in other hereditary conditions. In the DNA mismatch-repair genes, monoallelic pathogenic variants underlie Lynch syndrome (MIM #120435, ORPHA:144), whereas biallelic constitutional defects cause childhood-onset constitutional mismatch repair deficiency (CMMRD) with a distinct clinical spectrum (MIM #276300, ORPHA:252202) [14, 15]. In the homologous-recombination pathway, monoallelic variants in genes such as BRCA2 or PALB2 confer hereditary breast and ovarian cancer susceptibility (MIM #612555, ORPHA:145), while biallelic loss produces Fanconi anemia subtypes with early-onset hematologic and solid tumor risks (MIM #605724 and 610832, ORPHA:84) [16]. These single-gene examples demonstrate how the clinical and biological context may hinge on whether one or both alleles are affected, a principle that motivates careful attention to the cis/trans architecture, segregation, and potential multilocus genotypes in NF1. By contrast, a dual RASopathy diagnosis denotes two independent germline conditions in different Ras/mitogen-activated protein kinase (MAPK) pathway genes (for example, NF1 together with PTPN11 or KRAS). This is a multilocus scenario, not a single-gene allelic-status issue, and it requires rigorous variant curation and segregation to avoid over-attribution [17]. Here, we present an index patient with two NF1 variants and then synthesize primary literature on (i) cis doublets, (ii) germline trans configurations, (iii) biallelic somatic inactivation in café-au-lait macules (CALMs), tumors, and juvenile myelomonocytic leukemia (JMML), and (iv) bona fide dual RASopathy diagnoses, closing with a practical phasing/interpretation workflow. Case report A 10-year-old boy was referred for ongoing evaluation of NF1. Clinical features include multiple CALMs, bilateral Lisch nodules, plexiform neurofibromas, macrocephaly, focal areas of signal intensity on brain magnetic resonance imaging, and a low-grade glioma. Neurodevelopmental/behavioral history is notable for attention-deficit disorder, motor developmental delay with muscular hypotonia, reading and writing disability, and balance difficulties. On this basis he fulfills the clinical diagnostic criteria for NF1 [11]. Germline testing was performed outside our institutions at an external accredited laboratory using Sanger sequencing and MLPA on peripheral-blood DNA and identified two heterozygous NF1 variants [NM_001042492.3(NF1):c.3763 C > T (p.Gln1255*) and NM_001042492.3(NF1):c.3826 C > T (p.Arg1276*)], each classified as pathogenic. As expected for Sanger-based testing, quantitative variant allele fractions were not generated and read-backed phasing was not possible. Sensitivity for low-level mosaicism is limited and deep intronic or structural alterations beyond MLPA detection were not assessed in the work-up. A detailed three-generation history was non-diagnostic. On examination, both parents each had single CALM. Neither NF1 variant identified in the proband was detected in parental samples by targeted Sanger sequencing of blood DNA. Allelic phase has not been established. Given these constraints, two configurations were considered most plausible: (i) a cis doublet (both variants on the same allele), which is compatible with the child’s classic NF1 phenotype, or (ii) a germline variant plus a post-zygotic hematopoietic (blood-restricted) variant, which can appear as a second hit on blood testing despite a monoallelic germline architecture. In that setting, the putative somatic change often shows a lower or discordant variant allele fraction and is absent in non-hematologic tissue. Comparison with an alternative specimen (e.g., cultured skin fibroblasts or hair roots) would clarify tissue distribution. A third possibility, reflecting the high de novo mutation rate in NF1, is that both variants arose as independent de novo germline events. In view of the biological implausibility of constitutional biallelic NF1 loss, such double de novo changes would most plausibly reside in cis, consistent with clustered/multinucleotide events and with reports of distinct NF1 pathogenic variants occurring in first-degree relatives [18]. In our patient, follow-up phase-resolved analysis on a new peripheral-blood sample established that the two NF1 variants reside in cis. (Fig. 1) Observed allele fractions were compatible with heterozygosity, with no evidence for blood-restricted mosaicism. This result, together with the literature synthesis presented here, underpins the phasing/interpretation workflow included to guide similar cases in practice. Integrative Genomics Viewer (IGV) (https://igv.org/) visualization of forward (red) and reverse (blue) next generation sequence reads from NGS analysis. The two identified heterozygous variants NM_000267.3:c.3763 C > T; p.(Gln1255*) (altAF 0.4954, 54 of 109 reads) and c.3826 C > T; p.(Arg1276*) (altAF 0.4698, 70 of 149 reads) in exon 28 of the - NF1 gene are located on the same allele (in cis). One forward and one reverse read were highlighted as examples (red boxes) to show the localization of both variants on the same reads in both directions Monoallelic cis doublets in NF1 Reports of two pathogenic NF1 variants on the same allele (cis; monoallelic) are rare but well documented. The most convincing demonstrations establish phase by allele-specific PCR/cloning and RNA/complementary DNA analysis. In two independent families, Stella et al. identified double de novo, closely spaced NF1 variants on a single allele (for variant-level details and current interpretation please see Table 1). Both probands had classic NF1 without a reproducible increase in severity [19]. In a large pedigree, Hernández-Imaz et al. showed two truncating variants in NF1 occurring in cis and segregating with NF1 [20]. Terzi et al. described a child with two pathogenic NF1 variants and a mild phenotype [21]. While the frameshift variant c.5214dup (p.Thr1739fs) meets pathogenicity criteria, the missense variant c.2894T > A (p.Ile965Lys) remains a variant of uncertain significance (VUS)/likely pathogenic variant with conflicting ClinVar interpretation on current review. Together with the phenotype and subsequent literature, this constellation is most consistent with a cis configuration. More recently, Watanabe et al. reported two adjacent de novo missense variants in NF1 [NM_001042492.3(NF1):c.3586 C > T (p.Leu1196Phe) and NM_001042492.3(NF1):c.3590 C > T (p.Ala1197Val)] proven in cis—an instructive example of closely spaced multiple mutations that most plausibly arose from a single, complex mutational event rather than two independent hits [22]. In this context, cis pairs are best interpreted on the haplotype-level allele rather than treated as two independent lesions. Closely spaced substitutions can constitute a multi-nucleotide variant, for which the joint codon-level consequence may differ from the sum of single-variant annotations. Haplotype-aware consequence calling and phase-aware pipelines help avoid such misannotation, especially for clustered de novo changes [23, 24]. Analogous germline phenomena have been attributed to transient hypermutability and double-strand-breaks-associated processes (including a maternal/oocyte-aging component), with population studies estimating that a measurable fraction of de novo single nucleotide variants occur in such clusters [25, 26]. Table 1Germline variants in NF1 and other Ras-pathway genes with variant-level details and current interpretation Reference Gene Variants (as reported) Transcript reference Curated 1 Classification 2 ClinVar Variation ID gnomAD allele count 3 Segregation Interpretive note NF1 variants in cis Present report NF1 c.3763C>T, p.Gln1255*NM_001042492.3c.3763C>T, p.Gln1255*PATH404504Not observedde novoProven in cis by single read analysisc.3826C>T,p.Arg1276*NM_001042492.3c.3826C>T, p.Arg1276*PATH2375565:1613938de novoStella et al. 2022 Case 1 NF1 c.3198-1G>A, p.Asp1067Phefs*21NM_000267.3c.3198-1G>APATH2901595Not observedParents clinically unaffected, not available for testingProven in cis by segregation and cloning of PCR products Effect of splice site variant was verified on mRNA level as r.3198_3199delAGc.3295A>T, p.Lys1099*NM_000267.3c.3295A>T, p.Lys1099*LPATHNot availableNot observedStella et al. 2022 Case 2 NF1 c.2546del, p.Gly849GlufsTer29NM_000267.3c.2546del, p.Gly849fsPATH996424Not observedde novoProven in cis by cloning of PCR productsc.2548G>A, p.Val850MetNM_000267.3c.2548G>A, p.Val850MetVUS1318994Not observedde novoHernandez-Imaz et al. 2013 NF1 c.6792C>A, p.Tyr2264*NM_000267.2c.6855C>A, p.Tyr2285*PATH1850822:1613200Co-segregation within familyProven in cis by segregation within familyc.6799C>T, p.Gln2267*NM_000267.2c.6862C>T, p.Gln2288*LPATH /PATH1755477Not observedTerzi et al. 2012 NF1 c.2894T>A p. Ile965LysNot indicatedc.2894T>A p. Ile965LysConflicting VUS / LPATH947673Not observedInherited from fatherDe novo variant assumed to have occurred on the paternal allelec.5213insTNot indicatedc.5214dup, p.Thr1739fsLPATHNot availableNot observedde novoWatanabe et al. 2023 NF1 c.3586C>T, p.Leu1196PheNM_001042492.3c.3586C>T, p.Leu1196PheLPATH / PATH10361581:1614048de novoNF1 variants proven in cisby single read analysis; patient in addition had a pathogenic de novoGABBR1missense variant which likely contributed to the phenotypec.3590C>T, p.Ala1197ValNM_001042492.3c.3590C>T, p.Ala1197ValConflicting VUS / LPATH216401Not observedde novo NF1 variants in trans Paterra et al. 2022 Fam. 1 Bettinaglio et al. 2024 NF1 c.62T>A ,p.Leu21HisNM_001042492.3c.62T>A ,p.Leu21HisVUS1523219Not observedInherited from affected fatherProband carrying both variants reported to display a more severe phenotype than his affected father and brother who carried only the c.62T>A variantc.528T>A, p.Asp176GluNM_001042492.3c.528T>A, p.Asp176GluConflicting VUS / LBEN / BEN416736645:1612888Inherited from mother not meeting NF1 criteriaPaterra et al. 2022 Fam. 17 NF1 c.3314+2T>CNM_001042492.3c.3314+2T>CPATH1918167Not observedInherited from affected motherProband carrying both variants reported to display a more severe phenotype compared to the relatives carrying one of the NF1variantsc.7532C>T, p.Ala2511ValNM_001042492.3c.7595C>T, p.Ala2532ValBEN / LBEN41678994:1613328Inherited from father referred to as healthyFauth et al. 2009 NF1 c.3046T>C, p.Cys1016ArgNot indicatedc.3046T>C, p.Cys1016ArgPATH237543Not observedNot determinedVariants proven in transby analysis of human/mouse hybrid cell lines containing the different chromosomes 17c.8131_8133delGTT, p.2711delValNot indicatedc.8194_8196del p.Val2732delLBEN1575712Not observedNot determined Dual RASopathy diagnoses Bertola et al. 2005 NF1 c.2531A>G, p.Leu844ArgNot indicatedc.2531T>G, p.Leu844ArgPATH3731:1611918de novoNeurofibromatosis-Noonan syndrome PTPN11 c.1909A>G, p.Gln510ArgNot indicatedc.1529A>G, p.Gln510ArgLPATH / PATH1334514:1614226Inherited from father diagnosed with Noonan syndromeThiel et al. 2009 NF1 c.4661+1G>CNot indicatedc.4724+1G>CLPATH / PATH374Not observedInherited from mother affected by NF1The index patient’s phenotype was considered to result from additive effects of her NF1and PTPN11variants PTPN11 c.5C>T, p.Thr2IleNot indicated de novo LPATH / PATH133494:1528266de novoPrada et al. 2011 NF1 c.4288A>G, p.Asn1430AspNot indicatedc.4351A>G, p.Asn1451AspLPATH / PATH569136Not observedInherited from motherIndex patient reported to exhibit severe clinical presentation of a RASopathy PTPN11 c.922A>G, p.Asn308AspNot indicatedc.922A>G, p.Asn308AspPATH1332612:1612192Not determinedBaquedano Lobera et al. 2019 NF1 c.7756G>T p.E2586XNot indicatedc.7819T>G,p.Glu2607*LPATHNot availableNot observedde novoDual RASopathy (NF1, Noonan syndrome) KRAS c.40G>A,p.Val14lleNot indicatedc.40G>A,p.Val14llePATH125892:1612846de novo1Standardized nomenclature for the respective MANE transcript: NF1: NM_001042492.3; KRAS: NM_004985.5; PTPN11: NM_002834.52Classification of pathogenicity according ClinVar entry (if available) or by evaluation on the basis of criteria proposed by the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ACMG/AMP)3Global allele count in gnomAD 4.1.0 (January 2026) Germline variants in NF1 and other Ras-pathway genes with variant-level details and current interpretation 1Standardized nomenclature for the respective MANE transcript: NF1: NM_001042492.3; KRAS: NM_004985.5; PTPN11: NM_002834.5 2Classification of pathogenicity according ClinVar entry (if available) or by evaluation on the basis of criteria proposed by the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ACMG/AMP) 3Global allele count in gnomAD 4.1.0 (January 2026) Taken together, cis doublets appear to coincide with classic NF1 and do not consistently shift phenotype beyond the usual spectrum—though numbers are small and careful variant-level curation is still essential (e.g., rule out that one of the two is a VUS). Mechanistically, the following is expected: when cis changes render the transcript subject to nonsense-mediated decay or otherwise produce a null allele, a second pathogenic variant on the same allele cannot further reduce dosage beyond loss of that allele. Only in scenarios where both changes are missense (or otherwise expressed) on the same transcript, therefore producing a doubly mutated protein, is a theoretical additive effect conceivable; to date, convincing clinical aggravation from such cis missense doublets has not been demonstrated. Rare germline trans configurations True germline trans (compound-heterozygous) NF1 configurations are uncommon and tend to weaken upon close reappraisal. In line with this, true germline biallelic loss-of-function (LoF; trans) is biologically implausible: NF1 knockout mice die mid-gestation with cardiac/neural crest defects, supporting the notion that constitutive biallelic LoF is incompatible with prenatal survival [27]. When such configurations are discussed, it is typically in the context of spinal neurofibromatosis (SNF), a recognizable NF1 subset characterized by numerous bilateral spinal nerve-root neurofibromas and frequent familial accumulation [28]. In the largest SNF cohort, two unrelated probands were reported with two NF1 variants in trans [29, 30]. In Family 1, the additional variant reported in trans to the family´s established variant NM_001042492.3(NF1):c.62T > A (p.Leu21His) (now curated as VUS) was NF1 (NM_001042492.3):c.528T > A (p.Asp176Glu), inherited from a clinically unaffected parent. In Family 17, the established variant was NM_001042492.3(NF1):c.3314 + 2 T > C and the additional variant was NM_001042492.3(NF1):c.7595 C > T (p.Ala2532Val), inherited from a clinically not evaluated parent. Both additional alleles now have (likely) benign curations by multiple clinical laboratories, based on population frequency (gnomAD), ClinVar interpretations, and the criteria proposed by the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ACMG/AMP) (Table 1). In addition, the suggestion that compound carriers were more severely affected rests on qualitative observations rather than standardized phenotypic assessment. Fauth et al. additionally described an adult with sporadic SNF and malignant peripheral nerve sheath tumor (MPNST) who carried two variants in trans, c.3046T > C (p.Cys1016Arg) and a small in-frame deletion c.8194_8196del (p.Val2732del) [31]. Inheritance was not reported. Likewise, the small in-frame deletion lacks supportive functional/segregation data. Present-day submissions classify the variant as low-evidence/benign-leaning. Taken together, across published cases the putative “second” NF1 alleles do not constitute bona fide LoF. Most would now be classified as benign or, at most, of uncertain significance. Although subtle modifier effects of rare missense variation in trans cannot be excluded, there is no compelling evidence to reject the null hypothesis that these configurations reflect coincidental co-occurrence in a gene with substantial background missense variation. The apparent enrichment in SNF may likewise reflect ascertainment bias. Consistent with this, the literature provides no convincing examples of germline biallelic pathogenic NF1 loss, whereas biallelic inactivation in tumors/JMML arises via somatic second hits and is well established. Biallelic somatic inactivation of NF1 in café-au-lait macules, tumors, and juvenile myelomonocytic leukemia Mechanism of the second hit Loss of heterozygosity (LOH) or a second somatic mutation affecting the wild-type NF1 allele is the canonical driver of NF1-associated neoplasia (Knudson´s two-hit model). Across NF1-associated lesions, including cutaneous and plexiform neurofibromas, MPNST, pilocytic astrocytoma, and myeloid disease, clonal cells typically harbor biallelic NF1 inactivation, while the germline remains heterozygous [32]. Notably, even melanocytes in CALMs can carry a somatic second hit, underscoring that the two-hit paradigm extends to most or even all benign and malignant NF1-related focal lesions [33]. Multiple mechanisms of the second hit have been demonstrated in human NF1-related tumors: (i) Mitotic recombination leading to copy-neutral LOH (17q uniparental disomy, UPD): In neurofibromas, mitotic recombination that renders the Schwann-cell clone homozygous for the germline NF1 variant is a recurrent mechanism. Copy-neutral LOH on 17q via segmental uniparental isodisomy is likewise common in hematologic disease [32, 34–38]. (ii) Interstitial deletions/microdeletions at 17q11.2: focal deletions removing the remaining wild-type NF1 allele occur in a subset of tumors. In neurofibromas, distinct type-2 second-hit deletions have been catalogued [39]. (iii) Compound heterozygous inactivation by two distinct somatic variants: Tumors may acquire an independent pathogenic variant on the second allele in the absence of LOH [35]. (iv) Structural events: A single well-documented neurofibroma showed translocation-mediated disruption of both NF1 copies [40]. Collectively, these data support biallelic somatic loss as the rule in NF1 tumorigenesis [32]. JMML: an NF1-driven subtype with recurrent 17q events Children with clinical NF1 have a markedly elevated risk of JMML. Neurofibromin restrains RAS–MAPK in hematopoiesis. Haploinsufficiency confers Granulocyte/Macrophage Colony Stimulating Factor hypersensitivity and expands myelomonocytic progenitors, establishing a leukemogenic field that, in JMML, progresses via somatic second hits causing biallelic NF1 loss. In NF1-associated JMML, leukemic cells almost invariably achieve biallelic NF1 inactivation via 17q UPD leading to copy-neutral LOH, somatic NF1 deletions encompassing the NF1 locus on 17q11.2, or independent somatic NF1 mutations on the second allele [41]. In the most comprehensive contemporary series, 25 JMML patients with NF1 were analyzed: ten had 17q UPD resulting in copy-neutral LOH at NF1, five had microdeletion of one allele with a pathogenic NF1 variant on the other, and nine were compound-heterozygotes for NF1 variants. The same study interrogated a comparison group of 16 JMML cases without clinical NF1 and lacking canonical Ras-pathway drivers (PTPN11/KRAS/NRAS/CBL): eight had NF1 altered leukemias (three with microdeletions plus a hemizygous variant, three compound heterozygotes, and two simple heterozygotes) supporting routine NF1 analysis in JMML even when NF1 is not clinically apparent [41]. These findings extend earlier work showing that, in NF1-associated JMML, copy-neutral LOH at NF1 due to mitotic recombination on 17q is frequent. Among the remaining cases, biallelic inactivation typically arises through compound heterozygosity for NF1 variants or interstitial deletions, with no consistent genotype–phenotype differences across mechanisms [35]. They also align with broader hematologic data establishing copy-neutral LOH as a recurrent mechanism in myeloid malignancies and with the World Health Organization/International Consensus Classification 2022 classifications, which recognize JMML, NF1-associated, as a distinct molecularly defined category [42, 43]. Of note, secondary cooperating lesions (e.g., ASXL1 and/or SETBP1 mutations) are common in both NF1-JMML and NF1-negative/JMML-negative subsets, underscoring the value of a comprehensive genomic work-up [41]. Solid tumors and hyperproliferative lesions Across NF1-associated lesions, a second somatic inactivation event in NF1 is a prerequisite for clonal outgrowth, but it is not sufficient for malignant transformation. Benign neurofibromas, including plexiform neurofibromas, typically show biallelic NF1 inactivation in the Schwann-cell lineage. Progression toward malignancy requires additional, cooperating genetic and epigenetic alterations beyond the NF1 second hit [44, 45]. The evolution typically proceeds from plexiform neurofibroma to atypical neurofibromatous neoplasm of uncertain biologic potential (ANNUBP) and ultimately to MPNST. Along this continuum, recurrent cooperating alterations accumulate, most prominently CDKN2A/B copy-number loss at 9p21 and inactivation of the Polycomb Repressive Complex 2 via SUZ12 or EED LoF, with consequent loss of trimethylation of lysine 27 on histone H3. Additional events (e.g., TP53 mutation, broad copy-number remodeling) are frequent in high-grade disease. These cooperating events are rarely, if ever, present in ordinary neurofibromas, and underscore the stepwise nature of transformation [46–49]. Constituional NF1 microdeletions illustrate this logic clinically. Individuals with germline type-1 17q11.2 microdeletions have more numerous neurofibromas, earlier tumor onset, and a markedly elevated lifetime MPNST risk (≈ 16–26%) compared with patients harboring intragenic NF1 variants. Concomitant SUZ12 haploinsufficiency, due to co-deletion within the type-1 interval, has been implicated as a key modifier, consistent with the central role of PRC2 LoF in malignant progression [3, 4, 50, 51]. Within benign lesions, second-hit mechanisms include mitotic recombination leading to copy-neutral LOH (17q UPD) that renders the germline NF1 variant homozygous, segmental 17q11.2 deletions removing the wild-type allele, and, less commonly, compound heterozygous somatic inactivation by two distinct somatic variants in trans [39]. Recent integrative and single-cell studies reinforce this framework: PRC2 loss is recurrent in NF1-MPNST and closely linked to malignant progression. Pan-genomic surveys catalog NF1, CDKN2A, and PRC2 alterations as core lesions across stages, and consensus pathology recognizes ANNUBP as a bona fide precursor with molecular features intermediate between neurofibroma and MPNST. Together, these data support a multi-hit model in which NF1 biallelic NF1 inactivation is necessary but never sufficient for malignancy [52–54]. Bona fide dual RASopathy diagnoses Double diagnoses involving pathogenic variants in NF1 plus another RASopathy gene have been documented in a small number of individuals who show a blended phenotype, Noonan-spectrum traits (Noonan syndrome, MIM #163950, ORPHA:648) together with characteristic NF1 lesions. The most common situation pairs NF1 with PTPN11. In the index report, a girl with neurofibromatosis–Noonan features carried two independent pathogenic variants—de novo NF1 c.2531T > G (p.Leu844Arg) and a paternally inherited PTPN11c.1529 A > G (p.Gln510Arg)—providing the first molecular proof of co-occurrence of both disorders in one person [55]. In a subsequent family, an 18-month-old girl with typical Noonan features, multiple CALMs, and bilateral optic pathway gliomas harbored a de novo PTPN11 variant together with a maternally inherited NF1 splice variant in NF1 c.4724 + 1G > C. Relatives who carried only the NF1 variant had few CALMs as their sole NF1 sign consistent with an additive, blended presentation [56]. A third case described a male neonate with severe congenital heart disease, arrhythmia, hypertrophic cardiomyopathy, airway compression, and early death in whom pathogenic variants in both NF1 c.4351 A > G (p.Asn1451Asp) and PTPN11 c.922 A > G (p.Asn308Asp) were identified [57]. Although Ras/MAPK hypersignaling was proposed as a unifying mechanism, the fatal course may also be plausibly explained by Noonan-associated cardiac and lymphovascular complications. Mechanistically, such dual Ras/MAPK pathway variants are not equivalent to biallelic NF1 loss. In tissues that have not acquired a somatic second hit, one NF1 allele remains intact. A germline Noonan-spectrum gain-of-function variant (e.g., PTPN11) would be expected to modestly raise baseline Ras/MAPK tone rather than recapitulate complete NF1 loss. Conversely, once a tumor clone undergoes biallelic NF1 inactivation, pathway signaling is already deregulated. A concurrent germline Noonan-spectrum variant may increase phenotypic severity but is therefore not expected to be a major additional oncogenic driver, and there is currently no empirical evidence that tumor behavior is systematically intensified by the presence of a second RASopathy gene variant [58]. These considerations help delineate Neurofibromatosis–Noonan syndrome (NFNS; MIM #601321, ORPHA:638) from bona fide two-gene diagnoses. Most individuals labeled NFNS carry NF1 variants alone, and recognized genotype–phenotype enrichments account for Noonan-like features with a relative paucity of cutaneous/plexiform neurofibromas [59–61]. In clinical practice, a patient with NF1 and PTPN11 variants will often still be diagnosed as NFNS on phenotypic grounds. To distinguish NF1-only NFNS from true digenic cases molecular testing is necessary. However, a proportion of individuals labeled NF1-only NFNS may harbor undetected variants in Noonan-spectrum genes or regulatory/noncoding changes not captured by panel or exome tests. Where phenotype suggests a blended Ras/MAPK presentation but targeted testing is uninformative, genome sequencing (to access intronic/regulatory and structural variants), RNA sequencing from an informative tissue (to detect aberrant splicing/expression), and long-read sequencing (to resolve phasing and complex rearrangements) merit consideration [62]. Because gene–disease knowledge and curation standards evolve, periodic reanalysis of existing exome/genome sequencing data increases diagnostic yield and should be incorporated into counseling and follow-up plans [63, 64]. Beyond PTPN11, dual diagnoses have been reported with KRAS. A child harboring a truncating NF1 variant c.7819T > G (p.Glu2607*) together with KRAS c.40G > A (p.Val14Ile) displayed features within the upper severity range expected for this Noonan-associated KRAS allele, rather than a clearly novel or uniformly aggravated phenotype due to the combination [65]. Finally, co-occurrence of NF1 and SPRED1 pathogenic variants within the same individual appears exceptional. Many putative examples resolve to family-level co-segregation (different relatives) or to a SPRED1 VUS alongside a clearly pathogenic NF1 variant, underscoring the risk of over-calling dual diagnoses without robust pathogenicity and segregation data [66, 67]. As a differential diagnosis, however, SPRED1 remains important in pigment-predominant, tumor-negative presentations. Roughly ~ 1–2% of NF1-like clinic cohorts ultimately have Legius syndrome, and about ~ 8% of children with ≥ 6 café-au-lait macules and no other NF1 criteria harbor SPRED1 variants [68]. In summary, bona fide dual RASopathy diagnoses involving NF1 and a second gene are rare and best interpreted as additive/blended systemic phenotypes. Lethality is not the rule. Current data do not demonstrate that adding a germline Noonan-spectrum variant increases the oncogenic consequences beyond those driven by biallelic NF1 loss within lesions. For suspected dual cases, we advocate comprehensive RASopathy-panel or exome testing with ACMG/AMP-level curation of each variant and, where possible, segregation/trio data—while resisting VUS-driven over-interpretation. Outside RASopathies specifically, large exome cohorts find ~ 2–5% of solved cases carry multiple molecular diagnoses, indicating that dual diagnoses do occur but are uncommon, and careful curation is essential before concluding two syndromes [17, 69, 70]. Diagnostic implications Interpretation of reports listing two NF1 variants is shaped by several recurrent sources of error. Many individuals with Noonan-like features and NF1 harbor a single missense or in-frame NF1 variant (i.e., NFNS), rather than two pathogenic variants in both genes. Apparent biallelic NF1 in tumors or copy-neutral LOH of 17q in JMML reflects somatic evolution, not a germline trans state. Co-variation of NF1 and SPRED1 within families frequently involves a VUS in one of the two genes, arguing against inferring a dual diagnosis until independent pathogenicity is established. Mosaicism of NF1 adds further complexity, as low variant allele fractions in blood may indicate post-zygotic changes rather than a second germline allele. Finally, although many NF1 variants arise de novo and often on the paternal allele, evidence for a paternal-age effect remains limited and is best communicated cautiously [11, 71–73]. For individuals with clinical suspicion of NF1, we consider an NGS-based NF1 assay that captures both sequence variants and exon- or gene-level copy-number changes (via validated NGS-based CNV calling and/or MLPA) an appropriate first-line diagnostic approach, consistent with current molecular testing practice for NF1 [1]. In pigment-predominant presentations without detectable tumor manifestations, commonly encountered in children, and in the absence of an affected parent, we recommend upfront inclusion of SPRED1 to address the key differential diagnosis of Legius syndrome. Inclusion of LZTR1 and PTPN11 should be considered, ideally within a panel-based testing framework, to cover LZTR1-related multiple café-au-lait spots and Noonan syndrome with multiple lentigines. When the phenotype is broad or atypical for NF1, an extended panel or exome/genome-based analysis should encompass a wider range of differential diagnoses, including other RASopathies, CMMRD, and other conditions. Trio-based testing can be applied when it provides added value for assessment of de novo status or variant phasing. Targeted Sanger sequencing is best reserved for familial testing and/or orthogonal confirmation. Finally, we note practical assay considerations for mosaicism detection (read depth and variant allele fractions, and, when indicated, alternative tissues), RNA-based analyses for suspected splice effects, and a selected role for long-read sequencing in unresolved phasing or structural-variant cases. For unresolved NF1-only NFNS presentations or suspected dual RASopathy without a confirmed second locus, reanalysis of prior exome/genome sequencing at defined intervals (e.g., 12–24 months) is recommended, given consistent incremental yields with updated pipelines, annotations, and gene lists [74]. Where two pathogenic variants in different Ras/MAPK genes are confirmed (e.g., NF1 plus a Noonan-spectrum gene), clinical management typically integrates surveillance frameworks from both conditions (NF1-directed tumor surveillance together with Noonan-directed cardiology and growth/endocrine evaluation). When neoplasia is present, particularly JMML or enlarging/atypical plexiform neurofibromas, germline studies are complemented by tumor–normal analyses. Copy-neutral LOH/UPD (e.g., on single-nucleotide polymorphism arrays), CNVs (MLPA/next-generation sequencing), and targeted NF1 sequencing frequently document biallelic somatic inactivation. These findings clarify that the germline remains heterozygous. Recurrence risk therefore follows the germline variant rather than the tumor genotype. In JMML, identification of biallelic NF1 loss supports classification as NF1-associated JMML even when clinical NF1 features are subtle. In pigment-only lesions (e.g., CALMs), demonstration of a second hit reinforces the two-hit model without implying additional germline risk. In cases initially reported with two NF1 variants, analysis of tumor tissue helps prevent misattribution of somatic architecture to a germline trans configuration. Taken together, consistent application of ACMG/AMP classification to each variant, rigorous phasing, and, where tissue is available, tumor-focused investigation provide a coherent framework for interpretation (Fig. 2). On present evidence, lesion biology in NF1 is best explained by single-allele germline NF1 with somatic second hits as the proximate driver, whereas true germline trans pathogenic configurations, particularly in SNF, remain unsubstantiated. Interpreting (two) NF1 variants detected in blood: phasing and differential workflow Note: Haplotype-aware consequence calling is recommended when two cis variants fall within the same codon/region Concluding remarks Monoallelic cis doublets, including closely spaced, de novo pairs, are rare but well documented and generally tracked with classic NF1, with no reproducible phenotype escalation. Germline trans configurations reported in SNF weaken on reappraisal: the “second” alleles typically do not satisfy current pathogenicity standards, the perceived severity relies on qualitative impressions, and follow-up functional data are equivocal. By contrast, somatic second hits producing biallelic NF1 loss are ubiquitous in lesions from CALMs to JMML and MPNST. Bona fide dual RASopathy diagnoses exist, most convincingly NF1 plus PTPN11, and yield blended systemic phenotypes. However, there is no empirical evidence that a concurrent Noonan-spectrum variant further intensifies tumor behavior once a clone has lost both NF1 alleles. The practical corollary is simple: in putative “double” cases, prove phase, prove pathogenicity, and look to tumor tissue to avoid mistaking somatic architecture for germline biology. Acknowledgements We are grateful to Astrid K. Gnekow for her sustained support in establishing Neurofibromatosis clinical care and research at our center. We also thank our clinical colleagues and the participating family for their collaboration. Authors’ contributions DAT and MK had the idea for the article. DAT, MZ, and MK performed the literature search, data analysis, and data curation. DS performed sequencing and phase analysis. MK drafted the manuscript. DAT, DS, PV, MCF, TR, and MZ critically revised the work for important content. Funding This work was supported by Deutsche Krebshilfe for cancer-predisposition research (DKH 70115888), by the Deutsche Forschungsgemeinschaft for studies on double heterozygosity (KU 3764/3 − 1), and by intramural funds from University of Augsburg for cancer-predisposition research. DAT is supported by the Clinician Scientist Program of the University of Augsburg. The funders had no role in study design, data collection, data analysis, interpretation, manuscript preparation, or the decision to submit for publication. Data availability Not applicable. Declarations Ethics approval and consent to participate Written informed consent was obtained from the legal guardians. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Martin Zenker and Michaela Kuhlen contributed equally to this work. References Associated Data Data Availability Statement Not applicable.