Acute pancreatitis in HIV-seropositive patients: a case control study of 44 patients.

PURPOSE

To analyze whether pancreatitis presents differently in HIV-seropositive patients compared to the general population, and to evaluate the accuracy of classic predictors of pancreatitis severity in patients with HIV infection.

METHODS

A multiyear, multicenter, retrospective study of 44 consecutive patients with acute pancreatitis and HIV and 44 consecutive control patients with acute pancreatitis without HIV.

RESULTS

Of 939 hospitalized patients with HIV, 44 (4.7%) had acute pancreatitis, 27 of whom had AIDS. Pancreatitis presented with similar clinical findings in HIV patients and controls except that HIV patients had greater anemia, hypoalbuminemia, and leukopenia (Student's t-test P < 0.0001, < 0.0002, < 0.0001, respectively), and a higher incidence of fever, diarrhea, and hepatomegaly. These differences probably stem from HIV-related immunosuppression and malnourishment rather than pancreatitis. Patients with HIV had a higher frequency of medication-associated pancreatitis due to pancreatoxic medications used in HIV patients (18 versus 2 cases; odds ratio [OR] 14.54; Student's t-test P < 0.0001). They had a lower frequency of gallstone pancreatitis (2 versus 22 cases; OR 0.05; P < 0.0001). Patients with HIV had a higher frequency of a severe hospital course, defined as prolonged hospitalization or death in hospital (22 versus 12 cases; chi-square P < 0.05). The presence of AIDS or leukopenia in patients with HIV was strongly associated with a severe hospital course. The Ranson and modified Glasgow scales were poor predictors of disease severity in HIV patients (eg, Ranson scale sensitivity 41%; positive predictive value 53%; negative predictive value 52%). These scales' lack of markers of immunosuppression impeded their performance. The APACHE II scale, which contains markers of immunosuppression, was a moderately robust predictor of disease severity in HIV patients (sensitivity 73%; specificity 68%; positive predictive value 70%; negative predictive value 71%). All three scales predicted the disease severity in control patients well.

CONCLUSIONS

Pancreatitis presents similarly in HIV patients as in the general population with the following significant differences: a high frequency of medication-associated pancreatitis, a low frequency of gallstone pancreatitis, a high frequency of HIV-related causes (most commonly from HIV-related drugs), additional symptoms and signs due to underlying immunosuppression, and a more severe hospital course. The APACHE II system can be used to predict whether a patient with HIV and pancreatitis is at risk for prolonged hospitalization or death in hospital.