Study design and setting

This was a cross-sectional analytical laboratory-based study. The study was conducted in the Department of Pathology at Makerere University College of Health Science in Kampala, Uganda. The department is located within Mulago National Referral Hospital which is the only national referral hospital where there is also cancer treatment center (Uganda Cancer Institute) which provides oncology services including pediatric oncology services for Ugandans as well as patients from neighboring countries such as Democratic Republic of Congo (DRC), Rwanda, South Sudan, and Somalia.

Patients’ characteristics and recruitment process

We included in the analysis data of patients who were diagnosed previously with nephroblastoma from January 2012 to December 2015. Retrospective collection of the required clinical data of the patients and retrieval of the formalin fixed paraffin-embedded tissue blocks of the patients were done followed by histological re-evaluation of the previous diagnosis. All cases with complete clinical information and available FFPE tissue block of good quality were included in the analysis. But all cases with incomplete clinical data, missing FFPE tissue blocks, and spoilt FFPE tissue blocks by insects were all excluded from the analysis.

Sampling method

A total of 83 cases with confirmed histological diagnosis of nephroblastoma were included in the analysis. Sampling of the cases was done conveniently, and all cases that met the inclusion criteria were selected.

Hematoxylin and eosin staining

For re-evaluation of the previous histological diagnosis, all of the FFPE tissue blocks of the selected cases were sectioned at the thickness of 4 μm and stained with hematoxylin and eosin stains as previous [18]. Tumor staging as well as risk stratification of the cases was done based on the International Society of Paediatric Oncology (SIOP) [19].

Immunohistochemical staining for immunohistochemical expression of P53 protein

The FFPE tissue blocks were sectioned at 4 μm thickness and placed on poly-l-lysine-coated glass slides and placed on a hot plate for dewaxing for 10 min followed by being brought down to distilled water and then rinsed in phosphate tris buffer (PTB) solution. Endogenous peroxidase activity was blocked by placing the sections in 3% H2O2 for 5 min and rinsed 3 times in distilled water. The slides were then incubated with DO-7 antihuman p53 monoclonal primary antibody for 60 min at room temperature followed by two 5-min washes in PTB solution. The sections were incubated with biotinylated link anti-mouse immunoglobulin for 30 min and washed twice in PTB solution, followed by incubating with streptavidin–biotin-horseradish peroxidase complex (DAKO LSAB 2 system) for 30 min. The charged glass slides were washed in PTB solution for 45 min and stained in diaminobenzidine for visualization of antigen antibody binding. Finally, the sections were counterstained with hematoxylin for immunohistochemical evaluation. Considering tumor heterogeneity which is more likely to affect positivity of a given antibody, we stained at least four tissue blocks for every case.

Assessment of immunohistochemical expression of P53 protein

The slides were evaluated using light microscopy by two experienced pathologists independently, and scoring for positivity and intensity of immunostaining of the tumor cells was done according to a previous study [20]. In cases where the two pathologists disagreed, a third opinion from the third pathologist (tie breaker) was sought, and final decision was made by obtaining two similar interpretations. For assessment of p53 immunostaining, tumor cells with clearly brown reaction in the nuclei were counted by monitoring at least 1000 tumor cells from more than 5 high-power fields (HPFs) where positive cells were present at relatively uniform density, and the percentage was then calculated. The density score for the percentage of tumor cells with nuclear positivity was quantified as follows: 0:1–25%, 1:26–50%, 2:51–75%, and 3: > 75%. The numbers of immunopositive cells were counted, and the case was categorized as negative when none or only a few (< 5%) cells on the whole slide showed weak staining [17]. Then intensity of nuclear staining of the tumor cells was graded as follows: weak (light brown color staining), 1; moderate staining (moderate brown color staining), 2; and strong (dark brown color staining), 3. Density and intensity scores were combined and converted to a third score. All stains with an intensity and density score higher than 2 received a score of 1 (positive cases), and other cases with a combination of density and intensity score less than 2 were considered to be negative [20].

Data collection and research tool

Information regarding clinical data on age of the patients, sex, lag period, tumor size, tumor laterality, preoperative chemotherapy, and clinical presentation was extracted from the patients’ files as well as patients’ laboratory requisition forms. Also, histological components (anaplasia, blastemal, and epithelial), status for lymph node and surgical margins, and blood vessel and ureter involvement by tumor vessels were evaluated from the prepared FFPE tissue sections and recorded in the self-designed data collection form.

Data analysis

Data collected were edited and cleaned for missing data and other errors by running frequency tables and crosstabs. Analysis was performed using STATA version 12.0. All categorical variables were presented in frequencies and percentages, and continuous variables were summarized in mean ± standard deviation (SD) and also median (interquartile range (IQR)). Binary logistic regression analysis was performed to obtain the independent predictors of immunohistochemical expression of P53 protein. Only variables with p < 0.05 in univariate regression analysis were carried to multivariable regression analysis to determine factors that would independently predict the expression of P53 after adjusting for possible confounders. Adjusted odds ratios (AORs) were calculated at 95% confidence interval (CI) in multivariable regression analysis. A two-tailed p < 0.05 was considered statistically significant.