Pancreatitis: EuroSIDA finds low incidence, no link to use of specific antiretrovirals

An analysis of reports from nearly 10,000 European patients has found the overall incidence of pancreatitis to be low, at 1.27 cases per 1000 person-years. The study, an analysis of data collected from the EuroSIDA cohort since 2001, found pancreatitis to be more common at lower CD4 counts, but otherwise unaffected by antiretroviral use.

The study was published in the the January 2^nd edition of AIDS. Previous studies have found pancreatitis to occur much more frequently and to be associated with antiretroviral agents, particularly ddI (didanosine, Videx) and d4T (stavudine, Zerit); an editorial published in the same issue comments on the surprising EuroSIDA findings.

EuroSIDA is an ongoing, prospective, observational cohort study which has enrolled over 14,000 HIV-positive patients from 92 sites across Europe, plus Argentina and Israel. EuroSIDA began collecting data on pancreatitis in June 2001; the study population for this analysis included 9678 cohort participants followed from this point onward. Most were male (75%) and white (88%), with a homosexual/bisexual risk of HIV transmission (41%). At baseline, median age was 40 years, median CD4 cell count 415 cells/mm³, and median viral load 141 copies/ml; most patients (78%) had been receiving antiretroviral therapy (ART) for at least six months before baseline.

Pancreatitis events were classified as definitive acute, definitive chronic, or presumptive. "Definitive acute" events were those either diagnosed by a treating physician, or meeting all of the following criteria: one or more characteristic symptoms present, elevated pancreatic enzymes (lipase > 1.5 x upper limit of normal [ULN], amylase > 1.5 x ULN, or pancreas-type isoamylase > 1.5 x ULN), and at least one imaging diagnostic suggesting pancreatitis. "Definitive chronic" events required a treating physician's diagnosis, known previous episodes, or pancreatic calcification shown on diagnostic imaging. "Presumptive" events included the following:

• patients with amylase > 4 x ULN without other explanatory conditions,
• patients having two of the following: one or more characteristic symptoms present, elevated pancreatic enzymes (as above), or at least one imaging diagnostic suggesting pancreatitis, or
• cases in which definitive source documentation could not be obtained.

Cases not meeting these criteria were not considered.

Over a total of 33,742 person-years of follow-up, 43 cases of pancreatitis (34 definitive, 9 presumptive) were reported, for an incidence rate of 1.27 per 1000 person-years. Comparable incidence rates were found between differing ART regimens, regardless of whether they contained d4T and/or ddI, or the duration of exposure to the regimen, as summarised in the following table:

In multivariate analysis, CD4 cell count at baseline was the only factor to significantly affect risk of pancreatitis. The incidence rate per 1000 person-years was 1.93 for those with baseline CD4 cell counts below the median of 415 cell/mm³, and 0.64 for those above the median. Each 100 cell/mm³ increase in baseline CD4 cell count lowered the risk of pancreatitis by 22% (rate ratio [RR] 0.78, 95% confidence interval [CI] 0.66 – 0.93, p = 0.002).

These findings are surprising in light of other reports, which describe considerably higher rates in (largely North American) populations on ART, and which found significant associations with the use of ddI and d4T. (See, for instance, here.) In an editorial in the same issue of AIDS, Jeffrey Fessel and Leo B. Hurley comment that the EuroSIDA data are "interesting and quite surprising, particularly in the light of other reports… from as recently as 2001" – and the editors' own review of patient data from Kaiser Permanente in northern California – which found incidence rates roughly five times higher.

The reliability of the EuroSIDA report is not called into question: the editors describe it as a "well-conducted study". The researchers themselves report on statistical sensitivity analyses and quality control measures, including an annual audit of a randomly selected 10% of patients, which should minimise the risk of missing any clinical events. Nevertheless, the question remains as to why the EuroSIDA figures are four to fivefold lower than comparable North American findings.

The observed time frame may be one factor; we do not know whether the European data represent a decline from levels prior to 2001. However, this would not be explained by recent decreases in the use of d4T and ddI, and the investigators note that they did not observe any trends over time. There is also no data on diet, lipid levels, alcohol use, or hepatitis C prevalence or treatment, all of which might affect pancreatitis risk.

The investigators write that they "observed a low incidence of pancreatitis within the EuroSIDA study, [with] no evidence to suggest an increase over time in the years 2001 – 2006. Furthermore, there was no association between specific antiretroviral drugs, or combinations of antiretroviral drugs, associated with the development of pancreatitis."

As Fessel and Hurley state in their editorial, these findings are "discordant with experience from north America;" the "reasons for these unexpected findings are unclear and require further study."