Characteristics of patients with cervical pain originating from the lateral atlanto-axial joint (C1-C2) Authors: Victoria Schouela, Christopher Mares, Évelyne Théberge, Marie-Claude Bois, Samuel Côté, Laurence Roy-Bélanger, Isabelle Denis Published: 2026 ABSTRACT AbstractBackgroundThe lateral C1-C2 joint is an understudied source of upper cervical pain and is typically investigated only after negative work-up at C2-C3 and C3-C4. The clinical profile of patients with presumed C1-C2 pain remains poorly defined, and the relationship between diagnostic block response and medium-term corticosteroid benefit is unclear.ObjectiveTo characterize the clinical profile of patients with persistent upper cervical pain after negative C2-C3 and C3-C4 work-up who underwent C1-C2 intra-articular injection, and to identify factors associated with immediate diagnostic block response and with 3-month response to therapeutic intra-articular corticosteroid injection.MethodsIn this prospective study, fifty participants were recruited from a tertiary/quaternary physiatry clinic between 2022 and 2025. All had persistent upper cervical pain after negative C2-C3 and C3-C4 medial branch blocks. Participants underwent fluoroscopy-guided intra-articular C1-C2 injection with lidocaine and dexamethasone. A positive response at T0 was defined as ≥50% pain relief at 30 min. A positive response at T3 was defined as ≥50% pain reduction and/or ≥5-point improvement on the Neck Disability Index.ResultsForty-nine participants were analyzed at T0 and forty-eight at T3. At T0, 22/49 participants (44.9%) responded to the diagnostic block. Facial pain was associated with lower odds of response, whereas predominant neck pain or combined neck and head/face pain was associated with greater odds of response. At T3, 16/48 participants (33.3%) met responder criteria. Electric shock-type pain was associated with lower odds of response, whereas reported pain on touch was associated with greater odds of response. Response at T0 did not predict response at T3.ConclusionIn patients with persistent upper cervical pain after negative C2-C3 and C3-C4 work-up, lateral C1-C2 pain may be more likely in those with predominant neck pain or combined neck and head/face pain. Factors associated with immediate diagnostic response differed from those associated with 3-month therapeutic benefit, suggesting that diagnostic and therapeutic phenotypes may not be equivalent at this level.Keywords: Lateral atlantoaxial joint, Upper cervical pain, Cervicogenic headache, Intra-articular injection, Diagnostic block, Treatment response Introduction Background Neck pain is a leading cause of disability, imposing a significant burden on individuals, healthcare systems, and economies [1]. While 54%-60% of chronic neck pain arises from the zygapophyseal joints (z-joints) [2,3], identifying the specific cervical level(s) responsible for the pain remains challenging due to the considerable clinical overlap, particularly in the upper cervical region [4,5] and a low correlation rate between degenerative changes on imaging and neck pain [6]. As a result, diagnosis often relies on diagnostic blocks. Epidemiology Despite extensive research on neck pain, the upper cervical spine, particularly the lateral atlanto-axial (C1-C2) joint, remains relatively understudied. Cooper et al. [7] estimated that the C1-C2 joint accounts for less than 5% of neck pain cases, though this may be an underestimate due to the later inclusion of this joint in their protocol [7]. The same year, Yin and Bogduk suggested a prevalence of up to 9% [8]. In the specific context of cervicogenic headache, Govind and Bogduk (2022) demonstrated that while the majority of cervicogenic headaches originate from the C2-C3 z-joints (62%), 7% can be traced back to the C1-C2 joint [9]. Collectively, these estimates confirm that the C1-C2 joint is a non-negligible source of upper cervical pain, yet the clinical profile of affected patients remains poorly defined. Upper cervical pain patterns and diagnosis Currently, there is no gold standard for diagnosing C1-C2 joint pain. In their consensus practice guidelines, Hurley et al. (2022) concluded that no pathognomonic features exist [10]. Pain referral patterns of C1-C2, C2-C3, and C3-C4 greatly overlap, often affecting the upper neck, suboccipital region, vertex, eye, and superior forehead [7,11]. However, C2-C3 tend to spread more to the parietal region and is far less likely to refer to the ear or the vertex; thus, the presence of pain in these regions may favour C1-C2 involvement [7,12]. In addition, despite C3-C4 presenting with similar prevalence and pain pattern, it is rarely involved in isolation [7]. The predictive value of clinical assessment alone remains limited. Aprill et al. [13] found a positive predictive value (PPV) of only 60% when using a combination of occipital pain, suboccipital tenderness, tenderness at the C1 transverse process and restricted atlantoaxial rotation, and increased headache with this maneuver, while Narouze et al. [14] using similar clinical criteria, reported a PPV of 47%. These findings underscore that controlled diagnostic blocks remain essential to confirm the symptomatic level [10]. As C2-C3 and C3-C4 are more prevalent sources of upper cervical pain and are more readily accessible to medial branch blocks (MBB), clinical algorithms typically evaluate these levels first, leaving C1-C2 as a diagnosis of exclusion when the initial workup is negative [7]. However, the proportion of patients with negative C2-C3 and C3-C4 blocks who ultimately respond to a C1-C2 intra-articular block has not been systematically studied, and the clinical profile of this population remains poorly defined. Better characterization of these patients could help clinicians recognize C1-C2 pain earlier and select candidates for C1-C2 investigation more efficiently. Unlike the cervical levels below where radiofrequency neurotomy of the medial branches offers a validated therapeutic option following positive diagnostic blocks, no such procedure is currently available for C1-C2 [10]. Intra-articular corticosteroid (CS) injection therefore serves as both the diagnostic and primary interventional treatment at this level, yet its longer-term efficacy remains largely unexamined. Objectives The objective of this study was to characterize the clinical profile of the patients with persistent upper cervical pain after negative C2-C3 and C3-4 work-up who underwent C1-C2 intra-articular injection, and to identify factors associated with immediate diagnostic block response and with 3-month response to therapeutic intra-articular CS injection. Materials and methods Study design This prospective study was approved by the Centre de Recherche du Centre hospitalier de l'Université de Montréal (CHUM) ethics review board (CHUM; protocol 2023-10293). All participants provided written informed consent before enrollment. Fifty participants were recruited from a physiatry clinic at the CHUM, a tertiary and quaternary university center affiliated with the Université de Montréal, between 2022 and 2025. Participants Eligible participants had previously failed the standard interventional treatment of upper cervical pain according to International Pain and Spine Intervention Society (IPSIS) practice guidelines [15]. The C2-C3 and C3-C4 facet joints were excluded as the primary pain source using third occipital nerve (TON) and C3p and C4 MBB [16,17] (Fig. 1). MBBs were considered negative if pain relief was less than 75% at 30 min post-block; this threshold was selected in accordance with IPSIS guidelines to minimize false-positive rates [15]. Patients with negative MBBs became candidates for a C1-C2 intra-articular injection. Pre-study selection, recruitment algorithm and study time points. Inclusion and exclusion criteria The inclusion criteria were: 1) age ≥18 years old, 2) consultation for upper neck pain with or without headache, 3) minimum symptom duration of 6 months with failure of conservative treatment (non-steroidal anti-inflammatories, analgesics, and physical therapy [18]), 4) past cervical imaging (x-ray, computed tomography (CT) and/or magnetic resonance imaging), 5) pain intensity of ≥4/10 at rest or with activity on the numeric pain rating scale (NPRS), 6) upper extremity pain, if present, being non-predominant, 7) absence of signs of motor deficit in the upper limbs or of cervical myelopathy, and 8) elimination of C2-C3 and C3-C4 as pain sources using MBB, considered negative if pain relief was less than 75% at 30 min post-block [16,17] (Fig. 1). Potential participants were excluded from the study if they met any of the following criteria: 1) anatomic anomalies or pathologies compromising procedural safety, 2) medical condition or allergy that did not permit the safe injection of contrast, local anesthetic, or dexamethasone, 3) pregnancy or breastfeeding, 4) suspected or confirmed cervical condition that was either inflammatory, infectious, or neoplastic, 5) cervical CS injection in the last 3 months, 6) use of anticoagulants, 7) clinical presentation predominantly in the upper extremity or neuropathic, or 8) cognitive impairment or language barriers that limited the ability to fill out the questionnaires. Outcome measures Prior to the C1-C2 injection, patients completed a standardized questionnaire, the NPRS, and the Neck Disability Index (NDI). The questionnaire included: demographic information, duration of pain, current treatments, pain location, pain characteristics and associated symptoms. Participants were also asked to mark the areas they experienced pain using a pain map diagram produced by Cooper et al. [7], permission to use the pain map was granted by the publisher. Radiological degeneration of the C1-C2 joint was assessed opportunistically on intra-procedural fluoroscopic images obtained during C1-C2 injections. Although CT is the reference standard for grading atlantoaxial osteoarthritis, fluoroscopic images were the only imaging modality systematically available for all participants at the time of injection. A simplified ordinal severity scale adapted from the CT-based atlantoaxial facet osteoarthritis grading system described by Shimizu et al. [19] was used. This adaptation was limited to features identifiable on anteroposterior fluoroscopic views (e.g., joint space narrowing, osteophyte formation, subchondral sclerosis). Grading was performed by one physiatrist and subsequently reviewed by two additional physiatrists, with final classification established by consensus. To identify whether the C1-C2 lateral joint is the source of pain, fluoroscopy-guided intra-injection of the joint itself is considered the reference diagnostic and therapeutic procedure according to IPSIS [15]. These injections were performed by three experienced physiatrists trained in spinal injections, as per IPSIS guidelines, using a combination of steroid and local anesthetic: 0.5 mL of dexamethasone (10 mg/mL) and 0.5 mL of lidocaine 2%. Dexamethasone was used to assess the longer-term therapeutic impact of intra-articular injections. The pain intensity was evaluated using the NPRS immediately prior to the injection, 30 min post injection (T0), as well as 3 months later (T3). The joint block was considered positive if the patient experienced ≥50% relief of their pain using the NPRS (responder) at T0. This 50% threshold was chosen to increase sensitivity in this exploratory cohort. The literature generally shows that a reduction of at least 2 points is clinically meaningful, but a 50% calculated reduction is supported as a threshold for substantial pain relief, is commonly reported in chronic pain studies and, as such, is the reference used in this study [[20], [21], [22]]. Responders were subcategorized into two subgroups: subgroup one experienced ≥75% relief, while subgroup two experienced partial relief (50-74%). Those who had <50% relief were considered to have negative joint blocks (non-responders). At T3, a follow-up phone call was conducted to assess changes in function and pain using the NDI and NPRS, respectively. For the NDI, a change of at least 5 points on 50 (equivalent to a change of at least 10% on 100) is considered to reflect a clinically important difference [23]. A responder at T3 was defined as a participant with a reduction in pain of at least 50% and/or an improvement of at least 5 points on the NDI. Statistical analysis Descriptive statistics were used to summarize the baseline sociodemographic, clinical, and radiologic characteristics of the study population (Table 1). To characterize the clinical profile of this cohort and to explore factors associated with immediate diagnostic block response and 3-month CS response, baseline variables were compared between responders and non-responders at T0 and T3. Variables associated with response at T0 were interpreted as candidate features of immediate diagnostic block response, whereas variables associated with response at T3 were interpreted as candidate features of 3-month CS response. Continuous variables are presented as means and standard deviations, and categorical variables as frequencies and percentages. Table 1Sociodemographic description of study population.Table 1Age (years)(49) 52.7 ± 16.2SexM18 (36.7%)F31 (63.3%)EducationElementary/high school16 (32.7%)Collegial/university33 (67.3%)Current employment typeManual12 (24.5%)Administrative12 (24.5%)Other25 (51.0%)CNESSTa/SAAQbNo44 (89.8%)Yes5 (10.2%)SmokingNo42 (85.7%)Yes7 (14.3%)Time since pain onset (months)85.7 ± 108.1aCNESST: Commission des normes, de l'équité, de la santé et de la sécurité du travail - Quebec's provincial government organization that oversees labor standards, pay equity, and occupational health and safety.bSAAQ: Société de l'assurance automobile du Québec - Quebec government crown corporation responsible for licensing drivers, registering vehicles, and providing public auto insurance for bodily injury. Sociodemographic description of study population. CNESST: Commission des normes, de l'équité, de la santé et de la sécurité du travail - Quebec's provincial government organization that oversees labor standards, pay equity, and occupational health and safety. SAAQ: Société de l'assurance automobile du Québec - Quebec government crown corporation responsible for licensing drivers, registering vehicles, and providing public auto insurance for bodily injury. At T0, participants were classified as responders or non-responders to the diagnostic block according to the change in pain intensity from pre-injection to 30 min post-injection. Responders were defined as those demonstrating a reduction of at least 50% on the NPRS. Responder proportions were reported with 95% confidence intervals (CI). Baseline characteristics, including sociodemographic variables, time since pain onset, mechanism of injury, co-therapies, pain characteristics, pain location, associated symptoms, and degenerative changes, were compared between responders and non-responders at T0. Associations between categorical variables were assessed using Fisher's exact test for 2×2 tables and exact tests for larger contingency tables, with two-sided exact p-values. Variables of clinical interest were subsequently entered into multivariable logistic regression models to identify factors independently associated with response. Regression coefficients, odds ratios (ORs), 95% CIs, and p-values were reported. Overall logistic regression model significance was assessed using the omnibus likelihood-ratio chi-square test, whereas individual predictors were evaluated using Wald chi-square statistics. The same approach was used at T3 to compare responders and non-responders to the CS injection. Comparative analyses and multivariable logistic regression were then performed to identify factors associated with response at T3. The association between response status at T0 and T3 was assessed using cross-tabulation and the chi-square test. Using the pain map described by Cooper et al. [7], comparative analyses were also performed to compare pain distribution among T0 block responders and non-responders, as well as differences in pain distribution between the two groups (Fig. 2). Pain mapping of responders and non-responders at T0, reproduced with permission from the publisher of Cooper et al. [7]. Results Baseline characteristics Of the 50 participants who consented to participate in the study, 49 of them received the intra-articular C1-C2 injection at T0. Only one was lost to follow-up after having completed the standardized questionnaire and prior to receiving the injection. The mean age was 52.7 years (SD 16.2). Of the sample, 36.7% were male and 63.3% were female. Most participants had a college or university level of education (67.3%), were non-smokers (85.7%), and did not report an injury related to work or a motor vehicle accident (89.8%) (Table 1). Response to diagnostic block at T0 At T0, 22 of 49 participants (44.9%; 95% CI, 31.6%–58.8%) were classified as responders to the diagnostic block. Among the responders, 9 (40.9%) achieved a pain reduction of 50% to 74%, while 13 (59.1%) achieved a pain reduction of ≥75%. Expressed as proportions of the total sample, these correspond to 18.4% (95% CI, 9.5%–30.8%) and 26.5% (95% CI, 15.8%–40.0%), respectively (Table 2). Table 2C1-C2 joint block responders versus non-responders at T0.Table 2N%95.0% lower CL for the column % based on N95.0% upper CL for the column % based on NReduction in NPRS ≥50% pre-post blockNo (<50%)(Non-responder)2755.1%41.2%68.4%Yes (≥50%)(Responder)2244.9%31.6%58.8%Responder subgroups: reduction in NPRS pre-post block≥75%1326.5%15.8%40.0%50%-74%918.4%9.5%30.8%NPRS: numeric pain rating scale.CL: confidence limit. C1-C2 joint block responders versus non-responders at T0. NPRS: numeric pain rating scale. CL: confidence limit. No differences were observed between responders and non-responders at T0 (diagnostic block) with respect to: sociodemographic characteristics, presence of degenerative changes, use of co-therapies, time since pain onset, mechanism of injury, pain quality, or associated symptoms (Table 2 and Appendix 1). Overall, baseline sociodemographic and most clinical characteristics did not differ between groups, with exception of pain location which emerged as the main domain associated with diagnostic block response. A significant difference was observed for pain location. In univariable analyses, facial pain was more frequently reported among non-responders (p = 0.019). Although the broader distribution of pain predominance did not reach statistical significance in univariable analysis, a descriptive pattern was observed whereby responders more often reported predominant neck pain or combined neck and head/face pain, whereas non-responders more often reported predominant head/face pain. In multivariable logistic regression, which assessed whether these pain-location characteristics were independently associated with response, the overall model was statistically significant (χ2 = 12.505, df = 3, p = 0.006; Nagelkerke R2 = 0.312). After adjustment for the other pain-location variables in the model, presence of facial pain remained independently associated with lower odds of response to the diagnostic block (Odds Ratio (OR) 0.178, 95% CI 0.042–0.759, p = 0.020). Compared with participants whose predominant pain was located in the head/face, those with predominant pain in the neck (OR 6.769, 95% CI 1.104–41.500, p = 0.039) or in both the neck and head/face (OR 9.208, 95% CI 1.181–71.813, p = 0.034) had significantly greater odds of response (Table 3). Table 3Multivariable logistic regression analysis of factors associated with response to the diagnostic block at T0.Table 3VariableOR95% CIp-valueFacial pain present0.1780.042–0.7590.020Predominant pain in necka6.7691.104–41.5000.039Predominant pain in neck and head/facea9.2081.181–71.8130.034Note: n = 47. Omnibus model χ2 = 12.505, df = 3, p = 0.006; Nagelkerke R2 = 0.312.aReference category: predominant pain in head/face. Multivariable logistic regression analysis of factors associated with response to the diagnostic block at T0. Note: n = 47. Omnibus model χ2 = 12.505, df = 3, p = 0.006; Nagelkerke R2 = 0.312. Reference category: predominant pain in head/face. Response to CS injection at T3 At T3, 48 participants were included in the statistical analysis. One participant was excluded because comparison of pain outcomes using the NPRS was not possible due to analgesic use; however, this participant remained included in the T0 analysis, as that portion was unaffected. Of the 48 participants, 16 (33.3%; 95% CI, 21.3%–47.3%) were classified as responders to the CS injection. Responders at T3 were further divided into three subgroups: 6 (37.5%) met both pain and NDI criteria (subgroup A), 7 (43.8%) met the pain criterion only (subgroup B), and 3 (18.8%) met the NDI criterion only (subgroup C) (Table 4). No differences were found between responders and non-responders at T3 in terms of sociodemographic characteristics, presence of degenerative changes, time since pain onset, mechanism of injury, pain location, or associated symptoms. With respect to co-therapies, all participants who received massage therapy were non-responders, and this association reached statistical significance (p = 0.044) (Appendix 2). Table 4C1-C2 CS responders versus non-responders at T3.Table 4Response to CSN%95.0% lower CL for the column % based on N95.0% upper CL for the column % based on NNon-Responders3266.7%52.7%78.7%RespondersAll respondersNPRS reduction of ≥50% and/or improvement NDI ≥5 pointsa1633.3%21.3%47.3%Subgroup ANPRS reduction of ≥50% and improvement NDI ≥5 pointsa612.5%5.4%24.0%Subgroup BNPRS reduction of ≥50% and improvement NDI <5 pointsa714.6%6.8%26.5%Subgroup CNPRS reduction of <50% and improvement NDI ≥5 pointsa36.3%1.8%15.7%CL: confidence limit.CS: corticosteroid injection.a5 points is equivalent to 10% on 100. C1-C2 CS responders versus non-responders at T3. CL: confidence limit. CS: corticosteroid injection. 5 points is equivalent to 10% on 100. Regarding pain quality, univariable analyses showed the absence of electric shock-type pain was significantly more common among responders than non-responders (p = 0.013). In multivariable logistic regression, pain quality remained significantly associated with treatment response (χ2 = 11.647, df = 2, p = 0.003; Nagelkerke R2 = 0.312). After adjustment for other pain quality variables included in the model, electric shock-type pain was independently associated with lower odds of response (OR 0.078, 95% CI 0.013–0.479, p = 0.006), whereas the presence of reported pain on touch by the participants on the questionnaire was independently significantly associated with greater odds of response (OR 7.083, 95% CI 1.169–42.927, p = 0.033) (Table 5). Table 5Logistic regression analysis of factors associated with response to CS at T3.Table 5VariableOR95% CIp-valueElectric shock-type pain present0.0780.013–0.4790.006Pain on touch present7.0831.169–42.9270.033Note: n = 46. Omnibus model χ2 = 11.647, df = 2, p = 0.003; Nagelkerke R2 = 0.312.Odds ratios for electric shock-type pain and pain on touch are reported for presence versus absence. Logistic regression analysis of factors associated with response to CS at T3. Note: n = 46. Omnibus model χ2 = 11.647, df = 2, p = 0.003; Nagelkerke R2 = 0.312. Odds ratios for electric shock-type pain and pain on touch are reported for presence versus absence. Association between block T0 and CS T3 responders No association was found between response to the diagnostic block at T0 and response to CS injection at T3. Of the 22 participants who responded at T0, 7 (31.8%) were also classified as responders at T3, whereas 15 (68.2%) were non-responders at T3. Response at T3 was also not significantly influenced by the degree of response at T0. Among participants with a pain reduction of 75% or greater at T0, 30.8% were responders at T3, compared with 33.3% of those with a pain reduction of 50% to 74% at T0. In addition, 9 participants who were non-responders at T0 were classified as responders at T3. Exploratory subgroup analyses of imaging-detected degenerative changes among T0-only responders, T3-only responders, and participants who responded at both T0 and T3 did not reveal any clear trends or statistically significant associations. Pain map Fig. 2 illustrates the frequency of pain localization among responders (Fig. 2A), non-responders (Fig. 2B), and the difference in pain distribution prevalence between the two groups (Fig. 2C) at T0. Although no statistically significant difference in overall pain localization was identified, the maps suggest distinct distribution patterns. Pain among responders appeared to be more localized to the head and upper neck, whereas non-responders demonstrated a broader pain distribution. Overall, the pain distribution observed among block responders in this study appears broadly consistent with the pain map described by Cooper et al. [7]. Discussion Clinical relevance of overall findings This prospective study sought to better characterize patients with presumed lateral C1-C2 joint pain after standard upper cervical interventional work-up had excluded C2-C3 and C3-C4 facet joints as pain generators, and to identify factors associated with both immediate diagnostic block response and 3-month therapeutic response following intra-articular CS injection. Among patients with persistent upper cervical pain after negative C2-C3 and C3-C4 work-up, the present findings support continued consideration of the lateral C1-C2 joint as a potential pain source and suggests that the factors associated with immediate diagnostic block response may differ from those associated with 3-month CS benefit. The observed T0 response rate of 44.9% suggest that the C1-C2 joint may contribute to pain in a substantial minority of patients in this selected population, whereas the T3 response rate of 33.3% indicates that only a subset experienced meaningful medium-term improvement following intra-articular CS injection. Importantly, response at T3 was not restricted to participants who responded at T0, underscoring that immediate diagnostic response and medium-term CS benefit should not be assumed to be equivalent. This distinction is clinically relevant because a positive diagnostic response at C1-C2 should not necessarily be assumed to predict sustained therapeutic benefit. Clinical features associated with diagnostic block response Although prior studies have emphasized the overlap in referral patterns among upper cervical structures, the present results suggest that pain topography may still be clinically informative. Participants with predominant neck pain, or combined neck and head/face pain, were more likely to respond to the diagnostic block than those with predominantly head/face pain. While these findings do not establish pathognomonic features of lateral C1-C2 pain, they suggest that a neck-dominant or mixed neck and head/face presentation may be more consistent with this diagnosis than a predominantly head/face pattern. The pain-mapping findings support this interpretation as responders demonstrated a more localized pain pattern. Overall, the pain map pattern was broadly consistent with the referral map described by Cooper et al. [7]. The lack of statistical significance may be partly attributable to the relatively small sample size in relation to the large number of pain-mapping data points. Immediate diagnostic response versus 3-month therapeutic benefit The lack of association between immediate diagnostic response and 3-month clinical outcome was unexpected, and it suggests limited predictive value of the diagnostic block for response to steroid injection. Of the 22 participants who responded at T0, only 7 also responded at T3, while 9 participants who were non-responders at T0 were classified as responders at T3. In other words, short-term anesthetic relief and medium-term CS response do not appear to be equivalent. Immediate block response may reflect whether the joint is an active pain generator at the time of assessment, whereas 3-month outcome may depend on broader factors, including inflammatory burden, chronicity, coexisting pain generators, and pain sensitization. The 9 participants who were non-responders at T0 but responders at T3 may reflect a delayed anti-inflammatory effect of the CS or natural symptom fluctuation in a chronic pain population. These findings suggest that immediate diagnostic block responders and 3-month CS responders may represent two overlapping but distinct clinical profiles. For clinicians, the practical implication is that immediate block response and medium-term therapeutic benefit should be interpreted separately: the former may help support diagnostic suspicion of C1-C2 joint involvement, whereas the latter may reflect additional factors influencing treatment response. Further clarification of these potentially distinct responder profiles is needed to help clinicians identify which patients are more likely to have C1-C2-mediated pain and which are more likely to derive sustained benefit from intra-articular CS injection. These interpretations should also be considered in light of the chronicity of pain in our cohort, with a mean pain duration of 85.7 months (SD 108.1), indicating longstanding pain that likely complicates the clinical picture. In such a population, both diagnostic and therapeutic responses may be influenced by factors beyond the injected joint itself, including the psychosocial factors commonly associated with chronic pain. Clinically, a positive diagnostic block does not necessarily predict sustained benefit from CS injection, and conversely, a negative immediate response does not fully exclude later improvement. This lack of predictability is consistent with the current paucity of evidence to support the use of facet steroid injections [24]. Although discordance between imaging findings and clinical presentation is well established in cervical z-joint pain, we explored whether radiologic degeneration at C1-C2 might help distinguish patients who responded at different time points. In the present cohort, exploratory subgroup analyses of degenerative changes did not reveal any clear trends or statistically significant associations, suggesting that imaging-detected degeneration was not helpful in differentiating these responder groups. This does not exclude the possibility that other clinical or radiologic factors, or combinations of factors, may better distinguish patients with C1-C2-mediated pain from those most likely to derive sustained therapeutic benefit. Further research is therefore needed to better define these responder profiles and improve clinical suspicion, patient selection and counseling. Clinical features associated with 3-month response Pain quality appeared to be more informative than pain location for response at T3. Electric shock-type pain was associated with markedly lower odds of response, whereas reporting of pain on touch of the upper neck was associated with greater odds of response. These findings should not be interpreted as features that identify C1-C2 joint pain itself, rather, they may help characterize the subgroup of patients who were more likely to derive medium-term benefit from CS injection. One possible interpretation is that electric shock-type pain reflects a more neuropathic or non-nociceptive pain phenotype that is less likely to improve following an intra-articular CS intervention. In contrast, pain on touch may reflect a more peripheral nociceptive or mechanically irritable pain state. These observations remain preliminary, particularly given the modest sample size and wide confidence intervals, but they suggest that pain quality may be more relevant than pain location when estimating the likelihood of medium-term benefit. In clinical practice, these pain-quality features may therefore be more useful for estimating the likelihood of a 3-month CS benefit than for confirming the diagnostic suspicion of C1-C2 joint involvement. Comparison with previous literature The recently published retrospective case series by Kaufman et al. [25] provides a useful point of comparison. They reported that 19 of 52 patients, or 37%, achieved at least 50% pain reduction at 2 weeks following intra-articular lateral C1-C2 joint steroid injection [25]. In subgroup analyses, Kaufman et al. did not identify significant differences in outcome according to ASL criteria previously proposed by Levin et al., in 2024 [5] (ASL: age >65, superior cervical/suboccipital neck pain, and exacerbation of pain with lateral cervical spine rotation), STIR edema, or posterior headache, and longer-term outcomes were not available. Their findings, like ours, show that only a subset of patients derive meaningful benefit from C1-C2 injection, and that readily identifiable structural or clinical markers may not be sufficient to predict outcome. However, the present study extends this literature by incorporating both an immediate diagnostic assessment and a 3-month follow-up, thereby allowing distinction between predictors of diagnostic block response and predictors of therapeutic response. This may help explain why efforts to identify a single responder profile have been inconsistent: the phenotype associated with a diagnosed painful C1-C2 joint may not be identical to the therapeutic phenotype associated with CS benefit. Strengths, limitations and future directions This study has several strengths. It was prospective, evaluated a clinically relevant population seen in physiatry clinics associated with tertiary and quaternary university teaching hospitals, and included patients only after exclusion of C2-C3 and C3-C4 as likely pain generators using MBBs. It also assessed both immediate and 3-month outcomes, permitting a more nuanced analysis of diagnostic and therapeutic response. Several limitations should be acknowledged. First, the sample size was modest, and subgroup and regression analyses were based on relatively few outcome events, resulting in wide confidence intervals that limit the precision of the effect estimates. In addition, the large number of exploratory comparisons increases the possibility of chance findings; the observed associations should therefore be considered hypothesis-generating. Second, this was a single-arm study without a control group. Consequently, improvement at 3 months cannot be attributed solely to the CS injection, as natural symptom fluctuation, regression to the mean, co-interventions, and other factors may have contributed to the observed outcomes. Third, the diagnostic threshold was set at ≥50% pain relief, which was lower than the ≥75% threshold used for the preceding MBBs at C2–C3 and C3–C4. Although this threshold is consistent with IMMPACT recommendations and is commonly used in the literature [22], it may have increased sensitivity at the expense of specificity, potentially overestimating the proportion of participants whose pain originated from the C1–C2 joint. Participants also received 0.5 mL of 2% lidocaine mixed with dexamethasone, rather than local anesthetic alone, as would more typically be used for a purely diagnostic block. This may have influenced the response observed at T0 and should be considered when interpreting the immediate block findings. Furthermore, the T3 responder definition combined improvement in pain and disability. Although clinically relevant, this composite definition may have introduced heterogeneity within the responder group. Fourth, participants were recruited from a highly specialized tertiary- and quaternary-care hospital-based outpatient physiatry clinic, which may limit the generalizability of the findings. The longstanding pain experienced by participants may further limit their applicability to broader or less chronic clinical populations. Lastly, the grading approach used to assess C1–C2 joint degeneration was adapted from a CT-based system and applied to fluoroscopic images. It should therefore be interpreted as a pragmatic, study-specific classification rather than a formally validated fluoroscopic grading scale. Dedicated CT imaging of the atlantoaxial joints would have provided a more reliable assessment of degenerative changes and should be considered in future studies. Future research should focus on validating these findings in larger, preferably multicenter prospective cohorts. Studies designed to separate diagnostic anesthetic response from therapeutic CS response would be particularly valuable, as would the development of a clinical prediction model integrating pain distribution, pain quality, examination findings, and imaging variables. Further work is also needed to clarify the relationship between C1-C2 injection response and radiological findings such as edema or degenerative changes, and to determine the durability of benefit beyond 3 months, in order to improve patient selection for C1-C2 interventions. Conclusion This study suggests that lateral C1-C2 joint pain may be more likely in patients with predominant neck pain or combined neck and head/face pain than in those with predominantly head/face pain, and that the factors associated with immediate diagnostic response differ from the therapeutic phenotype associated with 3-month benefit. Together, these findings may help refine clinical suspicion and patient selection for C1-C2 joint interventions, but they require confirmation in larger prospective studies. Funding This study received institutional support from the Université de Montréal, which covered the cost of statistical consultation. No external funding was received. Declaration of competing interests The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article.