The prevalence, incidence and natural history of primary sclerosing cholangitis in an ethnically diverse population

Primary sclerosing cholangitis (PSC) is a chronic cholestatic liver disease of unknown etiology [1] characterized by classic findings of multiple segmental strictures in the intra- and extra-hepatic bile ducts secondary to inflammation and obliterative fibrosis. Less common, the small duct PSC subtype is characterized by a normal cholangiogram and a better prognosis [2]. Inflammatory bowel disease (IBD) is found in 65-90% of PSC cases and though usually classified as ulcerative colitis (UC), the IBD associated with PSC has a unique phenotype including pancolitis, ileitis, rectal sparing and is often asymptomatic [3, 4].

Both environmental and genetic risk factors have been associated with PSC susceptibility, in some cases being common to the IBD background, as in the case of smoking [5‐7]. Interestingly, the association of PSC with the human leukocyte antigen DR3-B8-A1 haplotype found in Northern European populations [8, 9] has not been reproduced in Italian or Brazilian PSC cohorts suggesting that genetic risks for PSC may vary between populations [10, 11]. In African-Americans patients listed for liver transplantation in the US, we have recently shown that PSC is associated with HLA-B8, but not HLA-DR3 suggesting a common genetic risk factor near the HLA-B locus [12].

The overall risk and clinical presentation of PSC also appears to vary between populations. Population-based studies in different geographical areas of primarily Northern European ancestry have demonstrated annual incidence rates ranging between 0.41 and 1.2 per 100,000 person-years [13‐18]. In populations of Asia, Southern Europe and Alaska, the prevalence of PSC appears much lower and with a lower frequency of IBD [19‐25]. In contrast, small case series suggested that IBD patients of African descent are at greater risk of PSC, [26, 27] although this has not been recapitulated in a larger study [28]. Our own analysis of liver transplant registrants in the US identified African-American race as a significant risk factor for PSC patients even after controlling for socioeconomic and other factors [12]. The aim of the present study was to determine whether the ethnic differences identified in the liver transplant registrant population extends to PSC patients in general by measuring the incidence, prevalence, natural history, and IBD co-morbidity in a large health maintenance organization consisting of over 3 million subjects representing an ethnically diverse population.

The results of our study, which to our knowledge is the largest population-based case-finding study on PSC, demonstrate a prevalence of 4.03 cases per 100,000 and an incidence of 0.41 per 100,000 person-years. Similar to other populations, the majority of cases are males and IBD, particularly UC, is associated with PSC and is more frequent in males compared to females. We also identified several variables associated with the major disease outcomes, i.e. death, liver transplantation and cholangiocarcinoma.

PSC is a rare cholestatic disorder of unknown etiology for which there is no effective therapy. Although the prevalence of the disease appears to vary widely in different ethnic and racial groups, most population-based studies of PSC have been in relatively homogeneous populations in terms of ethnic and racial composition [13‐17]. We sought to determine the prevalence and incidence of PSC in an ethnically and racially diverse population. The Northern California region is comprised of many different ethnic and racial groups with non-Hispanic whites making up only a slight majority of 51.1% based on 2000 US Census data. Because a strong association between PSC and IBD has always been observed to a lesser or greater degree and the risk of IBD varies between ethnic and racial groups, the prevalence of PSC might be similarly affected. Alternatively, the genetic basis of PSC appears to be independent of IBD genetic risk factors suggesting that there may not be a strict correlation between rates of IBD and PSC [9, 32].

Similar to other PSC cohorts, we observed a predominance of males with a mean age at diagnosis in the mid-forties. The frequency of IBD was similar to other studies and we identified a significantly lower frequency of IBD in women with PSC, which is consistent with the findings of other smaller studies [33, 34]. These rates should be considered a lower limit of the actual frequency. Some subjects did not have a record of a colonoscopy and often times the IBD of PSC is asymptomatic [3, 35, 36]. Whether the gender differences we observed are due to differences in true IBD prevalence or in rates of diagnosis warrant further investigation.

One of the most striking differences in our study compared to prior studies is the lower incidence and prevalence of PSC in this population compared to populations of Olmstead County, Alberta, Wales and Norway. This would seem to contradict the assertion by some authors that there is an increasing incidence of PSC diagnoses since the widespread use of endoscopic retrograde cholangiography and magnetic resonance cholangiography was introduced. One possible explanation for this discrepancy could be the result of under diagnosis leading to an underestimation of the true disease prevalence. PSC is a rare disease that may not be recognized by a clinician without particular skills or awareness. In addition, because there is not an effective therapy for PSC and liver test abnormalities are frequent in IBD, some clinicians may prefer not to confirm the diagnosis of PSC when it is suspected, particularly if an invasive test is needed. It is notable that our estimates of PSC prevalence and incidence are similar to those of another recent large study of a subset of the British population (total population of 2,027,909), which like the present study, was not based on one or a few hospitals with a particular interest in PSC [17].

Alternatively, this discrepancy might be due to a lower incidence of IBD, particularly UC, within Northern California. However, a recent study of the KP database determined that the incidence and prevalence of UC and CD is similar to or greater than that of Olmstead County, Manitoba and Europe [37]. Nevertheless, differences in the ethnic distribution between IBD and the total population have been noted[37, 38]. A prior study of a subset of this population in Oakland, CA reported that non-Hispanic whites made up only 64% of the total population but represented 80% of the IBD cases [38]. In contrast, Hispanics and Asian Americans comprised a disproportionately smaller percentage of the IBD cases compared to the total membership. In the more recent study of the total KP population, African Americans and Asians constituted 16% and 9% of the total IBD population, respectively, compared to 6% and 16% of the membership[37]. Further, other studies suggested that PSC is rare in Asia and that African Americans are at greater risk of developing extra-intestinal manifestation of IBD, including PSC [23, 24, 26‐28]. Ethnicity was documented in the medical record in fewer than half of the cases preventing us from drawing conclusions about the effect of ethnicity on the prevalence or incidence of PSC.

The strengths of our study include the use of comprehensive paper and electronic medical records to ascertain details of the PSC cases, the representative population of the northern California KP and the community care setting as opposed to a tertiary referral center. In contrast to the UK study, we were able to confirm each case of PSC based upon accepted diagnostic criteria. In addition, the size of the population allowed us to more accurately estimate the incidence and prevalence of a relatively rare disease. Therefore, these results are likely to be more representative of the total population.

Our study did have some weaknesses. The lack of ethnic and racial data in a large subset of patients limited our ability to address risk assessment or determination of prognostic factors for the natural course of disease dependent on race. We also note that not all patients with PSC had colonoscopies and therefore our estimates of IBD are likely underestimates of the actual rates. Finally, although this population is generally representative of the total population of Northern California, there is some skewing away from the upper and lower income groups.

We have reported on the largest population based study of PSC to date and the only one to include an ethnically diverse population. Although the common features of age at diagnosis, male predominance and presence of IBD are similar to other studies, we observed lower PSC prevalence and incidence rates than reported in most other populations. Further studies in this or other similar populations will be helpful in determining if there are significant differences in PSC susceptibility and clinical outcomes between ethnic groups.