We thank Dr. Afzal and colleagues for their detailed commentary on our article, “Pre-Diagnosis Alkaline Phosphatase and Antimitochondrial Antibody Positivity Vary by Race/Ethnicity Among Patients With Primary Biliary Cholangitis” 1. We appreciate the opportunity to clarify the goal of this analysis. Racial/ethnic disparities in PBC diagnosis and treatment remain a concern in the United States. In much of the literature, PBC has been described as most frequently occurring among white women—we hypothesized that such preconceptions regarding “typical” patients could impact clinical suspicion and follow-up testing that leads to timely diagnosis. Given that a large proportion of PBC cases in our sample are men and/or AAPI, Black, or Hispanic 2, 3, our goal was to describe the presentation of PBC biomarkers (ALP and AMA) in a sample that included many “atypical” patients: (1) to determine whether these varied by race/ethnicity prior to diagnosis; and (2) to evaluate whether there was evidence of disparities in the relationship between ALP elevation and timing of diagnosis. We would also like to offer some corrections. Dr. Akbar et al.'s letter states that we did not adjust for socioeconomic status or health insurance type. We agree that these are important considerations that may impact access to care. However, as noted in the Methods, our analysis of prediagnosis ALP trajectories used pairs matched by income category, among other variables. Although we did not match by health insurance type, as shown in Table 1, the proportion of patients seen in hepatology specialty clinics was similarly high (84%–87%) across racial/ethnic groups. As a result, we do not believe it likely that observed differences in AMA-positivity and ALP reflect differences in access to care. With regard to genetic ancestry data, we appreciate the reference to our previous publication regarding differences in UDCA response by race/ethnicity and our speculation on the possible role of underlying genetic differences 4. That manuscript, in fact, proceeded from previous work that described racial differences in rates of UDCA treatment receipt 2. We agree with the letter's authors that race and ethnicity are social constructs, with variable and tenuous relationships to genetic ancestry. However, in the United States, they are strongly associated with social determinants of health, including exposures that may increase risk of PBC, and also have real physiological manifestations. For example, self-identified Hispanic ethnicity was a significant variable in the model our team developed to predict response to UDCA treatment 5. While genetic data may be more appropriate for studies of disease mechanisms or treatment outcomes, our analysis was designed to investigate differences in presentation of biomarkers within the racial categories assigned by our current health system and society. Dr. Afzal and colleagues also raised a concern that patients received AMA testing only once. However, all AMA tests in the observation period prior to PBC diagnosis were included. Although it is true that some AMA-negative patients do develop antibodies over time (as was noted in the Discussion, 5% of previously AMA-negative patients became AMA-positive in the postdiagnosis observation period), we found no significant racial/ethnic differences in the proportion of patients who received AMA testing in the prediagnosis period of interest. As a result, we do not believe testing frequency influenced the significant racial differences in AMA-positivity we observed in our analysis. The letter writers are correct that the frequency and type of comorbid autoimmune conditions indeed differ by race/ethnicity—PBC-autoimmune hepatitis (AIH) overlap syndrome in particular may be more common among Black and Hispanic patients than non-Hispanic Whites. However, these conditions do not appear to alter AMA positivity; for example, patients with PBC-AIH generally present with positive AMA in addition to other autoantibodies 6. Moreover, we did not observe significant differences in ALP trends that would likely be attributable to systematic bias resulting from differences in the prevalence of comorbid liver disease among our matched pairs. Finally, Dr. Afzal and his colleagues raise the point that our analysis may not be generalizable to patients without health care access due to referral and selection bias. Although they suggest addressing this limitation with population-based approaches, a key challenge in rare disease research is that it is not feasible to identify a sufficiently large cohort using random population sampling nor are such diseases generally included in large representative datasets such as NHANES (we note that SEERS is a national cancer registry). The solution is to use case-based study designs, but an unavoidable limitation of such designs is that—by definition—diagnosed cases have some contact with a medical provider. This limitation is often addressed by efforts to diversify the sample pool. For example, our original Fibrotic Liver Disease (FOLD) cohort was identified from over 14.5 million individuals drawn from 11 health care systems that serve geographically, racially, and socioeconomically diverse populations across the United States 2, 3. We note that even with this large sample pool, only 4241 confirmed PBC patients were identified across the 12 year observation window. As a result, we do not believe population-based sampling methods are feasible for PBC research. Prevalence of PBC is rising in the United States 3 and worldwide 7, likely due to UDCA treatment allowing patients to live longer. Unfortunately, racial/ethnic disparities in UDCA treatment 4 and continuation 8 persist. We are encouraged by the increasing number of second-line treatment options available for the large proportion of patients who do not respond or cannot tolerate UDCA 9, including the recent approval of PPAR agonists 10 in the United States. Our future analyses will incorporate these treatments, including any racial/ethnic disparities in access and treatment response. Sincerely, Caines A, et al.
Caines et al. (Mon,) studied this question.