Study area and population
This was a case control study, done in paediatric and oncology wards of the three clinical centres in north-western Tanzania, namely Bugando Medical Centre (BMC), Sengerema and Shirati designated district hospitals in Mwanza and Mara regions, between September 2010 and April 2011. BMC is the zonal consultant and teaching hospital located in Mwanza city; it has a 900 bed capacity and it serves around 14 million people from 6 regions of the Lake Zone, namely Mwanza, Kagera, Shinyanga, Tabora, Mara and Kigoma. About five children with tumours are admitted monthly for investigation and treatment. Sengerema and Shirati DDH have approximately 300 and 200 bed capacities, respectively, and they each serve around 30 and 15 cases of childhood lymphomas every year.
Cases were enrolled if they were ≤15 years of age with a diagnosis of NHL and if consent from caretakers to participate in the study was obtained. Controls were included if they were admitted to the general paediatric wards with non-malignant conditions and were matched for age and sex with NHL cases. Children with malignancies that overlapped NHL, e.g. acute lymphoblastic leukaemia, and children with relapse of NHL were excluded from the study.
Sample size estimation
Sample size was calculated using OpenEpi software, version 2
, in which Fleiss formulae for case control studies was applied to determine the appropriate sample size for a power of 0.8 and significance level of 0.05. The minimum sample size for cases was 1/3 (33) and for controls was 2/3 (66) making the minimum sample size of 99.
Study clinical procedures
All consecutive NHL cases were recruited until the required sample size was attained, whereas simple randomization was done to select two controls per each case from all eligible patients. The investigator or trained research assistant either admitted the patient or reviewed the files of all eligible children within 2 days of admission and using a structured data collection tool, demographic data, clinical data and required investigation results were recorded.
To collect samples for cytological or histological diagnosis of NHL subtypes, fine needle aspiration of the tumour was carried out, using a needle of 25 or 27 gauge and a 20 ml disposable plastic syringe, under sedation with midazolam (0.1 mg/kg) and ketamine (1 mg/kg), or tissue biopsy of the tumour was carried out under general anaesthesia. In the case of an abdominal mass, an ultrasound scan was used to guide the needle into the tumour. Fine needle aspirate was smeared between two standard microscope slides with one slide immediately fixed with 95% alcohol and the other air dried and then transported in a slide carrier to the histopathology laboratory. Alcohol fixed smears were stained by Papanicolaus stain and air dried slides were prepared and stained with Giemsa.
Tissue biopsies were taken in the operating theatre, fixed immediately with 10% formalin and then transported to the pathology laboratory. Histology slides were stained with haematoxylin and eosin. At least two pathologists examined the slides under light microscopy and children with confirmed NHL diagnosis were enrolled in this study.
Lumbar puncture was performed in order to obtain cerebral spinal fluid (CSF) for cytological evaluation. A sterile 22, 38 mm gauge lumbar puncture needle (BD spinal needle) was inserted between the fourth and fifth lumbar vertebrae and the stylet withdrawn to allow about 5 ml of CSF to freely flow through the needle into the sterile tube which was then sent to the pathology laboratory to detect the presence or absence of cancer cells. Presence of cancer cells in the CSF indicated involvement of the CNS.
Bone marrow aspiration was performed, using a 13,5 cm gauge bone marrow needle (Jamshidi bone marrow aspiration needle), under sedation with midazolam and ketamine; 1, 2 ml of bone marrow were aspirated into the syringe and then smeared on 8 slides (4 fixed with alcohol and 4 without fixation) that were air dried and sent to the pathology laboratory for cytological analysis. Presence of lymphoma cells indicated bone marrow involvement.
NHL staging was done using the St. Jude’s staging of childhood NHL
Blood was drawn from the median cubital vein and collected in plain and EDTA bottles and then sent to the haematology and biochemistry laboratories for analysis of complete blood count and erythrocyte sedimentation rate, serum creatinine, alanine amino transferase and aspartate amino transferase.
The Tanzania national algorithm for HIV testing was followed to determine the HIV status of both cases and controls, using two antibody tests, the SD Bioline (SD Standard Diagnostics, Inc) and the Determine (Determine HIV 1/2 Serum/Plasma Assays); the second test was carried out in samples reactive to the first test. Those samples that were reactive to the first and second tests were considered to be HIV positive. Discordant results were subjected to a third test, Unigold (The Trinity Biotech PLC)
. Counselling on HIV testing was done and consent was sought prior to testing. Among HIV positive children, CD4 cell count was measured using Facscount (BD, San Jose, California, USA).
Blood collection and DNA elution
Approximately 50 μl of blood was spotted onto each circle of Whatman 903 filter paper and then dried at room temperature overnight. Dried Blood Spots (DBS) were stored in individual ziplock bags containing a desiccant, and sent to the laboratory of the Department of Surgery, Oncology and Gastroenterology, Section of Oncology and Immunology, University of Padova, Unit of Viral Oncology-IOV IRCCS, Italy. From each 50 μl DBS, three 3 mm-diameter circles, equivalent to 5 μl of whole blood each for a total of 15 μl whole blood, were used to extract DNA with the DNA Micro Kit (Qiagen, Hilden, Germany) and resuspended in 50 μl final volume.
To control the ability of the eluted DNA from DBS to be amplified, 5 μl of DNA from each sample were amplified for the human telomerase reverse transcriptase (hTERT) located in the 5p15.33 (Gene Bank accession: AF128893), employed as housekeeping gene. Amplification was carried out as previously described
. A quantitative method based on Multiplex Real-Time PCR assay was performed to quantify EBV type 1 and EBV type 2, using a primer pair for the EBNA2 gene
[36, 37] and the probes designed to discriminate between EBV type 1 and EBV type 2 ( EBV type 1: 5'-FAM-AAT CCT CCT ACC CTC TCT TTA TGC CAT GTG TGT-TAMRA-3'; EBV type 2: 5'-Cy5 TGG GCT GTT AGT AGG GT-BBQ-3')
. A standard reference curve was obtained by five-fold serial dilution of two amplicons, one for EBV type 1 and the other for EBV type 2, and amplification was carried out, as already detailed
. The multiplex assay showed a dynamic range from 5 to 2×105 copies. The results were expressed as EBV-DNA copies/ml. In those who had both EBV 1 and 2, viral levels were added and their total viral loads were analysed to investigate the association between NHL and EBV viral load.
Data were screened and edit checks were done to minimize data entry errors. Data were analysed with the STATA 11 software (College Station, Texas, USA). Categorical variables were summarized as percentages and were analysed by Chi-square or Fisher’s exact tests where appropriate. Continuous variables were summarized as mean (standard deviation) or median (range) where appropriate. EBV-DNA levels of cases and controls were compared using Wilcoxon rank sum (Mann–Whitney) test. The association between NHL and EBV was investigated using conditional logistic regression model. Odds ratios (OR) and 95% confidence intervals (CI) were determined by maximum likelihood estimation. The data were considered significant if the p-value was < 0.05.
Ethical approval was obtained from the Joint Bugando Medical Centre/Bugando University College of Health Sciences research and publication committee and guidelines on ethical requirement for conducting research and sample transportation outside the country were adhered to. Confidentiality was assured and the study did not interfere with the decision of the attending physician in case of absence of the investigator.