Background The human papillomavirus (HPV) is an important aetiological agent in cancer but its involvement in retinoblastomas (RBs) is controversial.
Methods 64 formalin-fixed paraffin-embedded tissue blocks and 19 fresh-frozen specimens were subjected to multiplex PCR using PGMY09/11 primers, HPV genotyping, non-isotopic in situ hybridisation and immunohistochemistry for pRb and p16INK4a.
Results 24% of RBs contained HPV DNA. 90% of HPV genotypes were of high-risk (HR) type and 10% were of intermediate-risk (IR) type. HR HPVs 45, 59, 68 and 52 were detected for the first time, as were IR HPVs 82 and 73. There was only one HPV 18-positive case. Interestingly, no low-risk genotypes were identified. Nine formalin-fixed paraffin-embedded HPV-positive cases showed nuclear HPV positivity by non-isotopic in situ hybridisation. Immunohistochemistry did not show pRb expression in 67% of cases. 34% expressed nuclear p16INK4a, of which 20 cases were also positive for HPV by multiplex PCR. A statistically significant association between HPV and pRb expression status was observed (p=0.0001).The association of HPV with p16INK4a expression was also statistically significant (p=0.0001).
Conclusions While the presence of HPV in a subset of RB was demonstrated, its role in carcinogenesis needs further elucidation.
- human papillomavirus
- in situ hybridisation
Statistics from Altmetric.com
Retinoblastoma (RB) is the most common primary ocular malignancy in children1 with an incidence of 1 in 15 000–30 000 live births, regardless of sex, race or geography.2 3 A retrospective analysis of all cases of RB presenting to our centre (a tertiary care cancer hospital) revealed only 4.5% of bilateral tumours (unpublished data). While bilateral RB constitutes 27% of cases in developed countries,4–6 it accounts for around 10% of all RBs in developing countries.5 This prompted a search for the increased incidence of sporadic unilateral RB in our population. The human papillomavirus (HPV) seemed a likely candidate given that the E7 protein of HPV inactivates unphosphorylated pRb.7
Reports to date on the association of HPV with RB have been contradictory. Studies in the Mexican and South American populations have reported high-risk (HR) HPV types in 28% and 82% of their patients with sporadic non-familial RB, respectively.8–10 By contrast, none of the 40 cases of sporadic RB showed HPV positivity in a North American population.11 Two studies on Asian Indian populations show 0%12 and 48%13 of HPV positivity, highlighting the piquancy of this association even in a single geographical area.
The present study addresses the association of HPV with unilateral RB using HPV detection methods: multiplex PCR using consensus primers (PGMY 09/11) followed by genotyping using HPV linear array and in situ detection of HPV by non-isotopic in situ hybridisation (NISH). Correlation with pRb and p16INK4a (a surrogate marker of HPV infection) was also done.14
Materials and methods
Formalin-fixed paraffin-embedded tissues (FFPE) from histopathologically confirmed cases of RB during the years 1994–2009 (n=64) were retrieved from the Department of Pathology, Kidwai Memorial Institute of Oncology. Data from the available clinical records were collected for analysis. This study was approved by the institutional ethical committee.
Fresh cases of RB (n=19) collected in-house and from various eye hospitals in Bangalore were also included in the study. The samples were collected aseptically in RNAlater prior to DNA extraction and stored at −80°C until use.
Representative sections were studied to assess differentiation. Differentiation in RB was defined as the presence of at least one of the following: Flexner–Wintersteiner rosettes, Homer Wright rosettes and fleurettes. Differentiation could be assessed in 75 cases which had adequate viable tissue.
DNA extraction from FFPE tissue (n=64) was done using the guanidium thiocyanate method.15 A new microtome blade was used for taking scrapes from each tissue block, with DNA extraction and PCR analysis being carried out in small batches with known positive and negative controls in every run. Care was taken to avoid cross contamination of samples during all steps of the procedure. Also, three separate rooms were used for various procedures. Five tissue scrapes of 5μ from each of the paraffin blocks were deparaffinised in xylene at 45°C for 2 h. The deparaffinised tissues were then treated with methanol and incubated in guanidium thiocyanate buffer (20 mM Tris pH 8.0; 1 M guanidium thiocyanate, 25 mM mercaptoethanol, 0.5% sarcocyl) containing proteinase K (6 mg/ml) at 55°C overnight. Proteinase K was inactivated by heating in a dry bath at 95°C for 10 min. DNA was then purified using the phenol: chloroform method. Commercial kits (DNeasy tissue DNA extraction columns, QIAGEN, Hilden, Germany) were used to extract DNA from fresh tumour tissue (n=19). Good-quality DNA was obtained in both FFPE and fresh tissue as checked by β-globin PCR.
Plasmid constructs of HPV obtained from across the globe and DNA extracted from squamous cell carcinoma of the cervix served as positive controls for the study. The negative controls were DNA extracted from (1) archived FFPE blocks of non-neoplastic eyes (n=5) obtained from fetal autopsy specimens and (2) prospectively collected non-neoplastic eyes (n=15) from an eye bank.
In-house multiplex PCR on DNA from FFPE and fresh tissues
DNA extracted from FFPE and fresh tissues was amplified for HPV by the in-house multiplex PCR comprising of primers towards the β-globin gene (PCO4 and GH2O)16 and HPV L1 consensus primers (PGMY09/11).17 The parameters for denaturation, annealing and elongation of the strands were as described earlier.17 Amplified products were detected by running the amplicons in 1.5% agarose gel in 1× TAE (Tris acetic acid EDTA) with ethidium bromide and visualised using a UV transilluminator.
Genotyping of HPV
Samples which were positive for HPV by multiplex PCR were genotyped using HPV linear array kit following manufacturer's instructions (Roche Molecular Systems, Branchburg, NJ, USA). Briefly, this comprised of two steps: (1) multiplex HPV PCR using the reagents provided in the kit and (2) genotyping of both low- and high-risk HPV types with the positive and negative controls provided in the kit.
Non-isotopic in situ hybridisation for detecting HPV in FFPE tissue sections
NISH was done on all HPV-positive cases which were positive by multiplex PCR. We standardised an in-house NISH using a cocktail of digoxigenin labelled HPV probes covering both high-risk and low-risk (LR) HPV genotypes (HPV 11, 16, 18 and 51) generated using digoxigenin 11-UTP (Roche Diagnostics, Mannheim, Germany) as per the manufacturer's instructions.18 The conditions for NISH were optimised in our laboratory as per the procedure of Han et al.19 Briefly, 5μ sections taken on silane-coated slides were rehydrated and treated with HCl and proteinase K, followed by RNase A and acetic acid (20%). After acclimatisation in the prehybridisation buffer for a short period, tissue DNA was denatured and hybridised at 45°C overnight with the denatured digoxigenin labelled probe cocktail. The next day, the slides were washed, blocked and incubated in appropriate dilution of anti-digoxigenin alkaline phosphatase for 1 h at 37°C. The signal was detected by incubation in substrate solution comprising of BCIP/NBT (Sigma Aldrich, St. Louis, MO, USA) overnight at room temperature (RT). Reactivity of the probe to each of the HPV types was checked by dot blot hybridisation using DNA from various plasmid constructs of LR and HR HPV genotypes (data not shown).
Immunohistochemistry for pRb (n=83)
Immunohistochemistry for pRb was done using a super sensitive polymer—HRP detection system kit (BioGenex, San Ramon, CA, USA)—following the manufacturer's instructions. Briefly, 5μ thick tissue sections on silane-coated slides were rehydrated; epitope retrieval was done by pressure cooking in 0.01 M citrate buffer at pH 6.0. Following quenching with peroxide block for 5–10 min at RT, the sections were treated with power block for 10 min at RT. Prediluted primary antibody to pRb (Clone 13A10, Novocastra, Newcastle, UK) was added and incubated for 1 h at RT. Secondary antibody conjugated with horseradish peroxidase polymer was added, followed by substrate solution (one drop of DAB chromogen mixed with 1 ml of substrate buffer) and haematoxylin counterstaining. The antibody used detected both hyperphosphorylated and hypophosphorylated pRb products. Sections incubated in normal saline instead of primary antibody comprised negative controls. Tissue sections from known positive cases (tonsil) were run with each batch.
Immunohistochemistry for p16INK4a (n=83)
Immunohistochemistry for p16INK4a was carried out using mouse monoclonal anti-human p16 antibodies [INK4] (Clone: G175-405, BD Pharmingen, San Diego, CA, USA). Briefly, 5μ paraffin sections were hydrated and quenched in methanol containing 0.1% H2O2. Epitope retrieval was performed in 0.01M citrate buffer (pH 6.0) by microwaving at 320 W for 30 min. The sections were blocked and incubated in primary antibody (1:20 in 1× PBS, pH 7.5) at 4°C overnight. The sections were then incubated in secondary antibody: biotinylated anti-mouse antibody (Vector Laboratories Inc., Burlingame, CA, USA) for 30 min at RT. Antigen–antibody complexes were detected with the streptavidin-peroxidase method (VECTASTAIN® ABC kit, Vector Laboratories Inc.) followed by incubation in the dark with 3, 3′-diaminobenzidine (DAB) containing 0.001% H2O2 in 1× PBS (pH 7.5). Tissue sections of carcinoma of the breast served as a positive control (as suggested by the manufacturers).
Evaluation of immunostaining
Immunoreactivity was considered significant when the characteristic immunostaining was observed in more than 10% of the cells. Batch to batch variation in staining intensity was compensated by including, each time, a positive control slide that consistently stained intensely for pRb and p16INK4a. The sections were evaluated blindly by the pathologist (RVK).
The χ2 test was used to analyse the association between various parameters. The correlation between various immunohistochemical markers was studied using Spearman's rank correlation. A p value less than or equal to 0.01 was considered to be statistically significant.
Clinical features and histopathology
The clinical features of the 83 cases are shown in tables 1 and 2. The mean age at diagnosis was 2.7 years. Leukocoria was the first sign of the disease in all cases. Differentiation could be assessed in 75 cases: 48 (64%) cases were differentiated and 27 (36%) were undifferentiated. HPV positivity was seen in 59% of the undifferentiated tumours, whereas only 8.3% of the differentiated tumours were positive for HPV. Further, this association between differentiation and HPV presence was statistically significant (p=0.0001). There was no significant difference between HPV-positive and HPV-negative tumours with regard to gender, laterality or age at presentation (p=0.2525).
Multiplex PCR and genotyping
HPV was detected in 20/83 (24%) cases of RB. Of these, 9 were from 64 archived cases (14%) and 11 from 19 fresh cases (58%). Nine genotypes were detected: seven HR (78%) and two intermediate risk (22%). Single genotypes were found in 13 cases (65%). The list of genotypes detected is given in table 3.
Non-isotopic in situ hybridisation
Nine samples of FFPE tissues (positive by HPV multiplex PCR) showed nuclear positivity of HPV within the tumour cells (figure 1B). The nuclear positivity in one case could not be assessed since there was little representative viable tumour in the tumour block available. Cervical squamous cell carcinoma served as the positive control (figure 1A). A section from a neoplastic eye (which was negative for HPV by PCR) was the negative control (figure 1C).
Nuclear staining of more than 10% of the tumour cells was considered as positive immunostaining. The staining of normal retina and endothelial cells within the tumour served as internal positive controls (figure 2B). pRb was expressed in 33% (27/83) of the cases (figure 2C). Of the HPV-positive tumours, 15/20 (75%) were pRb negative.
Nuclear and cytoplasmic staining for p16INK4a of more than 10% of the tumour cells was considered as positive staining (figure 2D). Thirty cases (36%) expressed p16INK4a, of which 20 were also positive for HPV by multiplex PCR. The remaining ten cases were consistently (twice) negative by PCR.
A statistically significant association between HPV and pRb expression status was observed (p=0.0001).The association of HPV with p16INK4a expression was also statistically significant (p=0.0001). There was an apparent inverse but statistically insignificant correlation between p16INK4a positivity and pRb negativity independent of the HPV status (r=−0.066, p=0.55).
Several papers from different parts of the globe, including one from India, have reported the presence of either only HR HPV in RB8 10 or predominantly LR types.9 20 In sharp contradiction, no HPV was found in two studies—North American11 and Asian Indian.12 This disparity inspired this project on the largest study population to date, that is, 83 cases of unilateral RB in which we found predominantly HR HPV in our population of Asian Indians.
HPV was detected in 20/83 cases (24%) of RB, of which 70% was the HR HPV 16 alone or in combination. Of the six HR HPV types (16, 18, 31, 33, 35 and 51) described in earlier studies, HPVs 16, 18 and 35 were detected. HR HPVs 45, 59, 68 and 52 were detected for the first time; as were the intermediate-risk HPVs 82 and 73. There was only one HPV 18-positive case and interestingly, no LR genotypes were identified. The wider spectrum of types detected could be attributed to the use of the PGMY 09/11 L1 primer pools for multiplex PCR which is capable of identifying 37 HPV genotypes. The chances of contamination were excluded stringently at every step and by including 20 non-neoplastic eyes (fetal or banked) which were consistently HPV negative. The higher frequency of HPV detection in fresh tissue (11/19) versus FFPE tissue (9/64) reflects the DNA degradation over time that could have taken place in disparately buffered formalin-fixed tissue. Significantly, histopathologically undifferentiated tumours harboured HPV. In situ hybridisation allows precise spatial localisation of target genomes in biologic specimens.21 To the best of our knowledge, this is the first study to correlate PCR findings with NISH. The presence of intracellular HPV DNA in nine archived tissue blocks clinches the presence of HPV in RB.
Immunohistochemical studies showed that 75% of HPV-positive RB (15/20) did not express pRb. This correlates well with the known pRb inactivating function of the HPV E7 oncoprotein of HR HPV types.22 In the remaining 25% of HPV-positive tumours where pRb was present, alternative pathways akin to pRb-independent HPV E7-mediated cervical carcinogenesis governed by oestradiol may be operative.23 pRb inhibits transcription of the cyclin-dependent kinase inhibitor gene p16INK4a. The negative feedback loop between pRb and p16INK4a would thus result in a reciprocal overexpression of the latter in HPV-positive tumours.24 All 20 HPV-positive tumours overexpressed p16INK4a as expected, underscoring its surrogate marker status.14 Levels of HPV that were too low for detection or presence of HPV types not amplified by the technique used might explain the finding of positive p16INK4a immunostaining in ten HPV-negative tumours.
As argued by Gillison et al11 the absence of HPV-positive RB in a North American group of patients may be partially explained by factors like diet and ethnicity that might modify the behaviour of HPV and contribute to the different results obtained in different geographical regions. This does not explain the absence of HPV in a group of 30 RBs from North India12 which shares a high-spice diet similar to the rest of India, Mexico and South America. Clearly, this warrants a multicentric study within the country and globally, to indisputably clarify the role of HPV in the aetiopathogenesis of sporadic RB.
In conclusion, we found HR HPV types in RB tumour which was proven using more than one technique. The level of expression of the tumour suppressor pRb, which is the main target of the HPV E7 oncogene, also revealed that the protein is not expressed in the majority of the HPV-positive cases. Lastly, the finding of intact pRb in a small proportion of HPV-positive cases warrants further study investigating alternative pathways in the oncogenesis of this tumour.
Professor Dr E M de Villiers, Dr Michel Favre, Dr A Lorincz, Dr W Lancaster and Dr T Matsukura kindly provided the HPV plasmid constructs for the study. We extend our sincere thanks to HBTR, NIMHANS and Lions Eye Bank, Bangalore, India, for providing FFPE blocks and fresh-frozen specimens of non-neoplastic eyes. We also thank Dr Sujatha, Dr Krishna R Murthy, Dr Hadi, Dr Ashwin Mallipatna and Dr Usha Kini for providing fresh samples/blocks for the study.
RSJ and RVK contributed equally to this work.
Funding This work was funded by the Indian Council of Medical Research, New Delhi, India (Project No: 5/13/47/05/NCDIII).
Competing interests None.
Ethics approval Ethics approval was obtained from the Institute Ethics Committee, Kidwai Memorial Institute of Oncology, Bangalore, Karnataka, India.
Provenance and peer review Not commissioned; externally peer reviewed.