Pancreatitis in pediatric human immunodeficiency virus infection Tracie L. Miller, MD, Harland S. Winter, MD, Lynn M. Luginbuhl, MD, E. John Orav, PhD, a n d K e n n e t h Mclntosh, MD From the Combined Program in Pediatric Gastroenterology and Nutrition and the Division of Infectious Diseases,Children's Hospital, the Department of Biostatistics, Harvard School of Public Health, Harvard Medical School, and Boston City Hospital, Boston, Massachusetts Because pancreatitis has been reported frequently in adults with human immun o d e f i c i e n c y virus infection, we sought to determine the i n c i d e n c e of pancreatitis in children with acquired i m m u n o d e f i c i e n c y syndrome by reviewing all records of children with AIDS, their serum amylase and lipase levels, and the factors associated with pancreatitis through a case-control analysis. During a 6-year period pancreatitis d e v e l o p e d in 9 (17%) of 53 pediatric patients with AIDS. Six children had vertical transmission of infection and three patients had a c q u i r e d HIV infection through c o n t a m i n a t e d b l o o d products. Pancreatitis dev e l o p e d at a median a g e of 5.2 years (range 1.2 to 20 years). All patients had vomiting and a b d o m i n a l pain. When the patients were first seen, lipase values were e l e v a t e d more than amylase values (p = 0.028). Amylase and llpase levels d e c l i n e d at c o m p a r a b l e rates. In the case-control analysis, p e n t a m i d i n e isethionate was significantly associated with pancreatitis (p = 0.02); the risk was greater in patients who r e c e i v e d p e n t a m i d i n e isethionate and had absolute CD4 T-lymphocyte counts less than 100 cells/mm 3 (p = 0.001). Infections associa t e d with the onset of pancreatitis included cytomegalovirus (4), Cryptosporidium (1), Pneumocystis carinii p n e u m o n i a (3), and Mycobacterium avium intracellulare (I). Coinfection with cytomegalovirus was associated with a protracted course in four children. Ultrasonographic examination demonstrated biliary ductal dilation 6 months after the onset of pancreatitis in one child. Seven children have died at a mean of 8 months after the initial onset of pancreatitis; the one living child has survived S months from the onset of pancreatitis. We c o n c l u d e that pancreatitis is c o m m o n in pediatric patients with AIDS and m a y be related to pentamidine isethionate exposure, especially when absolute CD4 T-lymphocyte counts are less than 100 cells/mm 3. Serum amylase levels do not always a c c u r a t e l y predict the onset of pancreatitis; serum lipase levels should be measured in children with symptoms. The onset of pancreatitis in an HIV-inf e c t e d child is a poor prognostic indicator. (J PEDIATR1992;120:223-7)

Chronic nonspecific diarrhea, wasting syndromes, and opportunistic infections of the upper and lower gastrointestiSupported in part by National Institutes of Health grants F32 DK 08518, 5UO1 AI 25934, and 5UO1 AI 27557. Submitted for publication July 8, i991; accepted Sept. 3, 1991. Reprint requests: Tracie L. Miller, MD, Combined Program in Pediatric Gastroenterology and Nutrition, Children's Hospital, 300 Longwood Ave., Boston, MA 02115. 9/20/33511

nal tract are among the most frequent complications seen in children with acquired immunodeficiency syndrome. 13 Pancreatitis has been recognized as one contributor to gasAIDS ddI HIV

Acquired immunodeficiencysyndrome 2' ,3 '-Dideoxyinosine Human immunodeficiencyvirus

trointestinal symptoms in adults infected with human immunodeficiency virus and has been linked to infection with

223

224

Miller et al.

such opportunistic microorganisms as cytomegalovirus, Mycoplasma tuberculosis, Toxoplasma gondii, and Cryptosporidium. 49 Additionally, neoplasms 8 and medications such as parenteral or aerosolized pentamidine isethionate,9-11 2' 3 '-dideoxyinosine,12-14 and sulfonamide-containing drugs 15 have been reported to cause pancreatitis in HIV-infected adults; HIV was implicated as the direct cause of pancreatitis in one patient. 16 Asymptomatic elevations in amylase levels without corresponding abnormalities in serum lipase values have been reported in some adult patients.17 Pancreatitis is not well characterized in children infected with HIV. In this series of patients we sought to determine the incidence and predisposing factors associated with pancreatitis in children with AIDS. METHODS

Patient population. The medical records of all 53 children with AIDS who were followed at Children's Hospital in Boston from 1986 to 1991 were reviewed in a retrospective manner for clinical and biochemical evidence of pancreatitis. Clinical pancreatitis was defined as an elevation in the serum lipase level of more than 2 SD from the mean, with accompanying symptoms of vomiting, abdominal distention, and intolerance to feedings. Medical records were reviewed for infection with opportunistic or other microorganisms, exposure to known toxic medications, total parenteral nutrition, absolute CD4 T-lymphocyte counts, and radiologic abnormalities at the time of diagnosis of pancreatitis. The weight of the patient at the onset of pancreatitis was used to determine nutritional status and was standardized by use of z scores. Renal function, as determined by blood urea nitrogen concentrations, serum creatinine values, results of urinalysis, and urinary output, was noted. Serum amylase and lipase levels at the time of diagnosis and at every subsequent measurement after diagnosis were recorded. Infection with HIV was documented by positive resuits on serologic tests for HIV both by enzyme-linked immunosorbent assay and by Western blot analysis, positive HIV culture, or both. For the case-control study, a comparison sample was drawn from the same population of patients who were determined to have AIDS by Centers for Disease Control criteria, 18 excluding those patients whose only feature of AIDS was lymphoid interstitial pneumonia. The control sample was matched by age, diagnosis of AIDS at matched age, and route of acquisition of HIV (vertical transmission vs contaminated blood products). Control patients had no clinical or biochemical evidence of pancreatitis. Statistical analysis. A linear regression was derived by means of the repeated-measures general linear model regression procedure (Statistical Analysis Systems software, SAS Institute Inc., Cary, N.C.) to document changes

The Journal of Pediatrics February 1992

in serum amylase and lipase activity with time. The declines in amylase and lipase activity after diagnosis of pancreatitis were exponential with respect to time, so the logs of both serum amylase and lipase values were taken after they were standardized to a percentage of normal values (i.e., each value for amylase was divided by 110 and each value for lipase was divided by 208). All available serial data on serum lipase and amylase values were obtained for each child and each episode of pancreatitis and were used in the linear-regression model. Data on one child with chronic pancreatitis were not analyzed because there were no changes in her clinical course that would have resulted in changes in either amylase or lipase values. A matched ease-control design was used to determine factors that may increase the risk of pancreatitis in the pediatric population with AIDS. Each patient with pancreatitis was matched to between one and four control patients of similar age with the diagnosis of AIDS and without pancreatitis. A random-effects analysis of variance model (SAS software) was used to determine the associations between absolute CD4 T-lymphocyte counts and weight z scores with pancreatitis in the index patient and matched control subjects. Association with pentamidine isethionate exposure and cytomegalovirus infection was determined by using the Cochran-Mantel-Haenszel procedure to derive stratified odds ratios with 95% confidence intervals and p values. RESULTS

Clinical observations. Nine children (six of them male) were found to have clinical and biochemical evidence of pancreatitis (Table I). All were symptomatically infected with HIV (Centers for Disease Control class P-2) and met the criteria for having AIDS. is These nine patients constituted 17% of all patients with AIDS (n = 53) who have been followed at Children's Hospital in Boston during the study interval. Pancreatitis was not diagnosed in other HIVinfected children without AIDS. The median age at onset of the first episode of pancreatitis was 5.2 years (range 1.2 to 20 years). The mean absolute CD4 T-lymphocyte count of the six patients in whom it was measured at the onset of pancreatitis was 10 ceils/ram 3. Two children had two episodes of panereatitis that were separated by intervals in which clinical symptoms abated and serum amylase and lipase values returned to normal. Four other children had a protracted course of pancreatitis, as demonstrated by persistently elevated serum lipase activity and continuing symptoms. Renal function was normal in all patients, as determined by blood urea nitrogen concentrations, serum creatinine values, results of urinalysis, and urinary output. Many of these children had active opportunistic or other non-HIV-related infections. Two children had disseminated cytomegalovirus infection defined by cytomegalovirus-as-

Volume 120 Number 2, Part 1

Pancreatitis in children with H I V infection

225

T a b l e I. Clinical characteristics of patients with A I D S and pancreatitis Patient No,

Age (yr)

HIV risk

Weight z score

1 2 3 4 5 6 7 8 9

1.2 1.4 1.6 2.0 5.2 8.0 12.7 16.3 19.8

Vertical Vertical Vertical Vertical Vertical Vertical Blood Blood Hemophilia

-3.29 -1.79 -3.47 -4.6 -1.28 - 1.11 -1.66 -1.55 -2.47

CD4*

N/A 10 N/A 39 4 5 1 N/A 5

CMV infection

Pentamidine exposure

Parenteral nutrition

Course of pancreatitis

Outcome (mo)

No Yes No Yes Yes'~ Yest No No No

Yes No Yes Yes Yes Yes Yes No Yes

Yes No No No Yes Yes Yes No No

Resolved Active at report Active at death Active at death Active at death Active at death Recurrent

11 (D) 5 (A) 0.5 (D) 5 (D) 3 (D) 10 (D) 7 (D) N/A 13 (D)

N/A Recurrent

CMV, Cytomegalovirus;D, dead; A, alive;N/A, follow-upinformation not available. *Absolute CD4 T-lymphocytecount in cells per cubic millimeter. tDisseminated infection with retinitis.

sociated retinitis in both children. Two other children were actively excreting cytomegalovirus in the urine within 1 week of the onset of symptomatic pancreatitis. The other five children had no evidence of cytomegalovirus infection or detectable antibody to cytomegalovirus. Both children with disseminated cytomegalovirus disease had active pancreatic disease until their deaths (3 and 10 months later). The course of their pancreatitis was not altered by ganciclovir therapy. One child with active pancreatitis was excreting cytomegalovirus at the time of death 7 months later. Six of nine patients had serologic evidence of past Epstein-Barr virus infection at the time of diagnosis of pancreatitis. Other infectious agents present at the onset of pancreatitis included Cryptosporidium and Mycobacterium avium intracellulare (each in one patient) and Pneurnocystis carinii (in three patients). Several children were receiving multiple drugs at the onset of pancreatitis (Table II). Five children were receiving pentamidine isethionate at the time they first had symptoms of pancreatitis; one patient was exposed to parenterally administered pentamidine isethionate 3 weeks before pancreatitis developed, and another was exposed to parenterally administered pentamidine isethionate 41/2months before the onset of symptoms. Five patients were receiving zidovudine at the onset of pancreatitis. None of the children was exposed to either ddI or 2 ' , 3 '-dideoxycytidine. Initial ultrasonographic examination of the abdomen revealed an enlarged echogenic pancreas in eight of the nine patients. One patient had biliary ductal dilation detected by ultrasonography several months after the onset of symptomatic, nonremitting pancreatitis. Results of endoscopic retrograde cholangiography in one child demonstrated normal pancreatic and biliary duct structure but revealed cytomegalovirus in the fluid obtained directly from the pancreatic duct as well as cytomegalovirus inclusions in tl-le duodenum. Seven of the eight patients who had pancreatitis and were

T a b l e II. Medications associated with pancreatitis m HIV-infected children

Medication Parenterally administered pentamidine isethionate Aerosolized pentamidine isethionate Zidovudine Trimethoprim-sulfamethoxazole

NO. of patients* (n = 9) 5 4 5 3

*Children were receivingmultiple medications at onset of symptoms.

being followed at Children's Hospital have died (88%). One patient's care was transferred to another state and no follow-up information was available. The mean survival time from the onset of pancreatitis was 8 months (range 2 weeks to 13 months). The remaining child was alive 5 months from the onset of clinical pancreatitis at the time of this report. Alterations in serum amylase and lipase levels. A t the onset of clinical symptoms, serum lipase levels were a mean of 16.4 times normal (3412 _+ 2966 U / L ) , whereas serum amylase levels were a mean of 3.9 times the normal level (427 _+ 391 m U / m l ; p = 0.028). One patient had normal serum amylase activity at diagnosis. Serum amylase values at the onset of clinical symptoms were consistently lower than serum lipase values were and returned to normal an average of 40 days earlier than lipase values did. The declines with time in serum amylase and lipase activities paralleled one another. Case-control analysis. There were no significant differences with respect to CD4 T-lymphocyte counts, coinfection with cytomegalovirus, or nutritional status (weight z scores) between patients with and those without pancreatitis after adjustments for age, diagnosis of A I D S , and route of H I V infection (Table III). There was a 4.7-fold increased risk of pancreatitis with pentamidine isethionate exposure (95% confidence interval for the odds ratio 1.02, 21.3, and this

226

Miller et al.

The Journal of Pediatrics February 1992

Table III. Case-control comparison of factors associated with pancreatitis in HIV-infected children Factor

CD4 count Weight z score Cytomegalovirus infection Pentarnidine isethionate Pentamidine isethionate and CD4 count

Pancreatitis in pediatric human immunodeficiency virus infection.

Because pancreatitis has been reported frequently in adults with human immunodeficiency virus infection, we sought to determine the incidence of pancr...
465KB Sizes 0 Downloads 0 Views