The critique, authored by Shubhendu Mohanty, Adarsh Jyoti Lakra, Hima Bindu Mantravadi, Prerna Uniyal and Dhanya Dedeepya, does not dispute the value of the underlying cohort. Assembling nearly a hundred well-characterized patients with two rare vasculitides, imaged to a common protocol and tracked over years, is genuinely difficult work, and the descriptive account of vascular distribution and treatment patterns stands as a useful contribution to the literature. The objection is narrower and more technical: it concerns what kind of question this particular study design is capable of answering, and whether the answer the authors reached was baked into the study before the first patient was enrolled.

The heart of the problem lies in the 2022 classification criteria from the American College of Rheumatology and the European Alliance of Associations for Rheumatology, which were used to allocate patients to the two diagnostic groups. Classification criteria are, by design, tools built to discriminate between named diseases. They are constructed from clinical and imaging features known to separate the conditions, and they are validated for that discriminating purpose. When a study then compares the groups on those same features and reports the differences as findings, the analysis becomes circular. The classifier and the outcome are the same variable, and the result is a foregone conclusion dressed in the language of discovery.

Age offers the clearest illustration of this circularity. The Takayasu criteria require disease onset at or below 60 years of age, while the giant cell arteritis criteria require patients to be 50 or older. The Italian study reported a median age of 33 in the Takayasu group against 76 in the giant cell arteritis group, with a p-value below 0.001. But as the commentators point out, this striking difference simply restates the entry rule. No patient over 60 could appear in the Takayasu group, and few under 50 could appear in the giant cell arteritis group. Reporting the age gap as a distinguishing feature of the two diseases adds no new biological information; it merely echoes the arithmetic of the classification scheme.

Vascular distribution presents a more consequential version of the same problem. The study’s methods state explicitly that the differentiation of cranial giant cell arteritis from its large-vessel form, and from Takayasu arteritis, was based on angiographic assessment. Yet the results then compare angiographic vascular distribution between the groups and report involvement of the axillary arteries, the aortic arch, the mesenteric vessels and the renal arteries as findings that distinguish the diseases. Polymyalgia rheumatica, which was present in 37.8 percent of the giant cell arteritis group and in none of the Takayasu group, falls into the same category, because it is a recognized component of the giant cell arteritis phenotype that informs the diagnosis in the first place. In each case, the feature that appears to separate the groups is the feature that was used to separate them.

Crucially, the commentators are not arguing that Takayasu arteritis and giant cell arteritis are the same disease. Their claim is epistemological rather than clinical: a cohort assembled through classification criteria cannot adjudicate whether the two entities are distinct, because the criteria were engineered to make them distinct. The features that could genuinely settle the question are precisely the ones the criteria do not use, and the Italian study actually measured several of them. Erythrocyte sedimentation rate and C-reactive protein, the two classic inflammatory markers, showed no significant difference between the groups, with p-values of 0.722 and 0.448 respectively. Neurologic and pulmonary manifestations likewise did not differ, and the long-term remission rate was reported as not significantly different in the abstract. An analysis restricted to these criteria-independent variables, the commentators suggest, would constitute a real test of the distinctness question, and would arguably make a more interesting paper.

The critique raises a second, independent statistical concern: the sheer number of comparisons. Across Tables 1, 2 and 3, the original study performed roughly thirty, eleven and ten statistical tests respectively, all evaluated at a significance threshold of 0.05 with no adjustment for multiple testing. At that volume of testing, chance alone guarantees that two or three findings will cross the significance threshold even if no true differences exist. Several of the differences that survived into the paper’s conclusion were marginal: dermatologic involvement at p equals 0.04, chronic liver disease at p equals 0.02, axillary artery involvement at p equals 0.02, and low-dose glucocorticoid use at p equals 0.04. The commentators note that Fisher’s exact test, which the authors appropriately chose for the sparse data, validates each individual test but does nothing to address the multiplicity problem. Their proposed remedy is straightforward: designate a small number of prespecified comparisons as primary, and label everything else exploratory.

A third concern involves confounding by referral pattern, a problem the original authors themselves acknowledged in their limitations section. One of the three participating centers is a regional tertiary referral center for Takayasu arteritis, which the authors offered as an explanation for the imbalance in group sizes. But the consequences of that arrangement extend well beyond group size. A referral center concentrates severe and refractory disease, so the Takayasu group was enriched for exactly the patients most likely to receive aggressive treatment. The study reported high-dose glucocorticoid use in 59.3 percent of Takayasu patients against 18.9 percent of giant cell arteritis patients, combination therapy with glucocorticoids plus immunosuppressants in 86.4 percent against 51.4 percent, and interventional vascular procedures in 25.4 percent against 5.4 percent. These figures, the commentators argue, reflect differences between two differently recruited populations at least as much as differences between two diseases, and the conclusion that the conditions respond differently to therapy does not follow from them.

The comment also catalogs a series of numerical inconsistencies that the authors are asked to reconcile at proof stage. The abstract reports the p-value for time from symptom onset to diagnosis as 0.025, while Table 1 and the results text give 0.029. The caption to Table 3 describes a cohort of 84 patients, although the study comprises 96. The subclavian artery row of Table 2 lists the giant cell arteritis proportion as 35140, presumably a typographical error for 35.14 percent. And the proportion of patients followed for at least 60 months appears as 72 percent in the abstract and discussion but 75 percent in the methods. None of these discrepancies is individually decisive, but together they underscore the importance of careful proofreading in studies that will inform clinical classification.

The broader lesson extends well beyond this single cohort. Classification criteria are indispensable tools for enrolling homogeneous patient groups into research studies and for standardizing clinical communication, but they are diagnostic heuristics, not ground truth. When researchers compare groups defined by such criteria on the variables that constitute them, they risk converting a definitional artifact into a purported biological finding. The commentators close with a constructive proposal: the Italian cohort remains a valuable resource, and a comparison restricted to criteria-independent features, governed by a prespecified analysis plan, would allow the data to speak to the question the study set out to answer. Until such an analysis is performed, the claim that Takayasu arteritis and giant cell arteritis require fundamentally different approaches remains plausible, but unproven by this evidence, a distinction that matters for the clinicians treating these rare and potentially devastating diseases of the large arteries.

Subject of Research: Methodological critique of a cross-sectional comparison of Takayasu arteritis and giant cell arteritis

Article Title: Comment on “Takayasu Arteritis and Giant Cell Arteritis: Results From a Cross‐Sectional Study of 96 Italian Patients”

Article References: Mohanty, S., Lakra, A. J., Mantravadi, H. B., Uniyal, P., & Dedeepya, D. (2026). Comment on “Takayasu Arteritis and Giant Cell Arteritis: Results From a Cross‐Sectional Study of 96 Italian Patients”. Immunity, Inflammation and Disease, 14(9), Article e70513. https://doi.org/10.1002/iid3.70513

Image Credits: AI Generated

DOI: 10.1002/iid3.70513

Keywords: Takayasu arteritis, giant cell arteritis, vasculitis, classification criteria, ACR/EULAR 2022, circular reasoning, multiple testing, referral bias, rheumatology, statistical methodology, large-vessel vasculitis, epidemiology