Full details of the methods are provided elsewhere [1, 2]. Briefly between 1994 and 1998, we recruited adults 15 years or older with a new diagnosis of cancer from the wards and out-patient clinics of the main hospitals in Kampala, Uganda. After informed consent and counselling, patients were interviewed and tested for infection with HIV-1 using the Cambridge Bioscience Recombigen ELISA (Cambridge, MA) on sera or the GACELISA method (Murex, Dartford, UK) on saliva. Cancer diagnoses were established by histology or other laboratory investigation, where possible. Diagnoses made on clinical grounds alone were reviewed by the investigators. The study was approved by the Committee on Human Research (VA Medical Centre and University of California San Francisco) and by the Uganda National Council for Science and Technology.
Following HIV testing, remaining sera were stored at minus 80 Celsius and were later shipped on dry ice to the Centres for Disease Control and Prevention, Atlanta, USA, for H. pylori testing. Assays were performed by a single investigator who was blind to the diagnosis of the patient from whom the blood was obtained. Briefly, H. pylori organisms were grown overnight in brucella broth (GIBCO Laboratories, Madison, WS) with 10% fetal bovine serum (Sigma, St. Louis, MO), 5 μg/ml trimethoprim and 10 μg/ml vancomycin (Sigma). H. pylori antigen extraction and protein isolation were done by gentle freeze-thaw sonication (Heat System, Farmingdale, NY) [8, 9]. A standard protein assay (Pierce, Rockford, IL) was used to determine the accurate and reproducible quantity of solid-phase antigen for our microtitre research ELISA . Cross-reactivity and specificity of H. pylori whole-cell antigens has been described previously [9, 10]. Optical density (OD) values at a wavelength of 492 nm were determined in triplicate for each biopsy-confirmed control patient sera, using a standard 96-well microtiter plate ELISA spectrophotometer (Fisher Scientific, Pittsburgh, PA). The mean OD values were then calculated. The ELISA cut-off values were derived using known H. pylori-positive and negative control sera as previously described [9, 11]. In previous validation studies the assay has demonstrated a high and reproducible sensitivity and specificity in African patients as compared to upper endoscopy and biopsy; sensitivity >88%, specificity >90% [9, 11].
Serological results were available for 50 people with non-malignant manifestations of HIV disease, recruited from the out-patient department of Mulago hospital and for 804 patients with cancer or benign tumours, for whom a stored blood sample was available for testing. The latter group comprised people with cancers of the oral cavity (26), oesophagus (38), stomach (21), liver (52), skin (22), breast (69), cervix (190), ovary (22), prostate (10), penis (14), eye (63), and non-Hodgkin's lymphoma (46), Hodgkin's disease (24), Kaposi's sarcoma (46), other cancer sites or types (126) and benign tumours (35).
Data were computerised by trained clerks using EPI-INFO software (CDC, Atlanta) and statistical analyses were conducted using STATA (STATA Corporation, Texas). Only a small proportion of those tested were seronegative for antibodies against H. pylori (optical density <0.9) or had an indeterminate result (optical density 0.9–1.3). In all analyses, those with indeterminate results were considered to be seronegative. In order to examine potential confounding factors, the risk of being seropositive for antibodies against H. pylori was examined in relation to various social and demographic factors among all patients combined (but excluding stomach cancer, which has been associated with H. pylori infection). Odds ratios (OR) were estimated using unconditional logistic regression modelling with adjustment for sex, age group (<30, 30–45, 46+) and HIV serostatus. When calculating odds ratios in relation to anti-H. pylori antibodies, for each cancer site or type, the comparison group included all other patients with the exception of stomach cancer. Tests for association used the χ2 test for linear trend on one degree of freedom and all p values are 2-sided. Risk factors for high titres of antibodies against H. pylori were examined amongst all patients combined (excluding stomach cancer), but no clear associations were identified and the data are not shown.