Human Papillomavirus Typing of Rare Cervical Carcinomas

Human papillomavirus (HPV) is causally associated with cervical carcinoma,1 the second most common cancer in women worldwide.2 Morbidity and mortality statistics vary from region to region. Countries with established screening programs, such as the United States, have low incidence rates, whereas neighboring Mexico has one of the highest incidence rates of cervical cancer in the world.3 

High-risk HPV types are so called because they are associated with cervical carcinoma. More than 40 distinct HPV types are known to infect the genital tract,4 the majority of which are cancer-associated. These high-risk HPV types can vary with geography and ethnic group,2 but HPV-16 is the predominant type found in women with cervical dysplasia5 and is overall the most common HPV type detected in cervical carcinoma.2 Certain histologic types of cervical carcinomas, such as adenocarcinoma,6,7 as well as rare variants of adenocarcinoma8 may have lesser association with HPV than squamous cell carcinoma. For epidemiologic study and vaccine development, it is imperative to determine the specific HPV types that are found in a wide variety of cervical carcinomas, especially those from high-incidence regions. Human papillomavirus typing may also play an important role in clinical outcome, as some types may be more virulent. In addition, certain histologic types of cervical tumors appear more closely related to particular HPV types, for example, adenocarcinoma and HPV-18.7,9 Although it is estimated that approximately 20% to 30% of HPV-infected women harbor multiple HPV types,5 little is known about HPV coinfection in cervical neoplasia. All of these issues need to be clearly delineated before we can fully understand the pathogenesis of cervical carcinoma.

Diagnostically, HPV typing should help in the determination of HPV persistence, the major risk factor for cancer development.7,10,11 Currently, an HPV DNA detection system, Hybrid Capture 2 (Digene Corporation, Gaithersburg, Md) is approved by the Food and Drug Administration for use in patient care. This assay uses signal amplification technology with RNA multiprobe cocktails to distinguish between high-risk and low-risk HPV types, but does not identify the specific HPV DNA type detected. For research purposes, most laboratories use a polymerase chain reaction (PCR)–based system to amplify a general or consensus region, for example, L1 of HPV.12–15 Typing is then based on either amplicon size, restriction enzyme digestion pattern,16 hybridization to type-specific probes,14,17 sequencing,18 or amplification using type-specific primers.15,19 Although some of these techniques are highly labor-intensive, others such as hybridization detection using probe-impregnated strips are relatively easy, rapid, and adaptable to multiple specimens. One of these, referred to as the line-blot assay, developed by Patti Gravitt in collaboration with Roche Molecular Systems,12,20 uses modified consensus primers (PGMY09/11) with 27 type- specific probes for genital HPV types. This technique has been used to determine the differences in the natural history of low-risk and high-risk HPVs,3,4 as well as the likelihood of HPV-16 infection predetermining risk for coinfection.5 

In this study, we histologically characterized and performed HPV typing on a diverse group of 43 human cervical carcinomas archived at the Institute of Cancer in Mexico City, Mexico.

Forty-three paraffin-embedded tissue blocks, archived at the Instituto Nacional de Cancerologia in Mexico City, were selected for their unusual histologies and sent to Louisiana State University Health Sciences Center at Shreveport (LSUHSC-S). These specimens were encountered during a period of 6 years (1995– 2000) during routine sign-out of surgical pathology specimens. Demographic data were not collected on these women, allowing for an exemption for informed consent by the LSUHSC-S Internal Review Board for Human Research.

Two pathologists working independently and blinded as to the HPV typing result reviewed the hematoxylin-eosin–stained slides. Discrepant results were resolved by reassessment of diagnostic criteria and consensus evaluation.

Five-micrometer sections cut from paraffin blocks were deparaffinized. DNA extraction was performed in duplicate (10 sections per specimen) using the Gentra Systems Puregene DNA Isolation Kit (Minneapolis, Minn), according to the manufacturer's recommendations for paraffin-embedded tissue with some modification.

All reagents for HPV amplification and typing were kindly provided by Roche Molecular Systems, Inc (Alameda, Calif). Human papillomavirus PCR and typing was performed using the Roche HPV Consensus PCR/Line Blot Genotyping reagents according to the method described.12,20 Controls included C33A (HPV negative), HeLa (HPV-18 positive), and SiHa and CaSki (both HPV-16 positive) cell extracts. Risk category for HPV types is as described elsewhere.3 

Rare and common tumors were compared for HPV types using the Fisher exact test.

Tissue histology results are given in the Table for the entire set of 43 human cervical carcinomas. Tissue histology was organized into 2 broad categories for comparison analysis. Common types included squamous cell carcinoma (n = 12) and adenocarcinoma (n = 2). Rare tumor types are defined as histologies representing less than 5% of all cervical carcinomas and included adenosquamous (n = 4), papillary (n = 2), villoglandular (n = 1), anaplastic (n = 10), transitional cell (n = 2), spindle (n = 1), adenoid basal (n = 5), colloid (n =1), neuroendocrine (n = 2), and glassy cell (n = 1) carcinomas.

Human papillomavirus typing results by histologic category for the 43 HPV-positive tumors are shown in the Table. Tissues with dual infections are noted and both HPV types detected are indicated. No significant difference was found between the presence of more than 1 HPV type (21% of both groups), positivity for HPV-16 (66% of rare tumors and 71% of common tumors), or absence of HPV types 16 or 18, although the rare cancers had a greater tendency toward more unusual types (28% of rare tumors and 7% of common tumors had no HPV-16 or -18 DNA). Human papillomavirus types other than HPV-16 and HPV-18 found in the rare category included HPV-52 (n = 6), HPV-84 (n = 2) (previously known as HPV- MM821), HPV-26 (n = 2), HPV-35 (n = 2), and HPV-58 (n = 1). Only 1 tumor in the common category lacked HPV- 16 or HPV-18 DNA. Non-HPV 16/18 types found in the common tumors included HPV-39 (n = 1) and HPV-31 (n = 1), both types found also in the rare category, and 2 types (HPV-33 [n = 1] and HPV-39 [n = 1]) not detected among the rare tumors.

Human papillomavirus 16 alone was the most commonly detected HPV type, found in 15 (51%) of 29 carcinomas in the rare category and in 8 (57%) of the 14 common cervical tumors. Human papillomavirus 18 alone was detected in 0% and 3 (21%) of 14 of the rare and common categories, respectively.

We analyzed 43 cases of cervical carcinoma from Mexico City. Twenty-nine of these cases involved unusual (rare) tumors. The 2 predominant tumors within this group of rare tumor types were anaplastic (n = 10) and adenoid basal carcinomas (n = 5).

While it still may be debatable as to whether cervical carcinoma in the absence of HPV etiology exists, it is assumed that this would be a rare instance.2 Our HPV-positive results for rare histologic types of cervical tumors confirm the broad scope of HPV-mediated oncogenesis.

In our study, we found no specific correlation between HPV type and tumor histology, albeit our numbers were small. Human papillomavirus 16 DNA, either alone or in combination with another HPV type, was detected in 66% of all rare and 71% of all common tumors. Of the 2 broad categories of tumors, the rare histologic cases tended to harbor more unusual HPV types than the common category, although this difference was not statistically significant.

Human papillomavirus 52 was detected in none of the common tumors, but was found in 6 (21%) of 29 of the rare types. Other non–HPV-16/18 types were found in less than 10% of the cases. The predominance of HPV-16 was not unexpected. In a large international study of HPV epidemiology performed in 1995 with tumors from 1035 patients,2 HPV-16 accounted for 52% of the types identified and was the predominant type in all countries except Indonesia, where HPV-18 accounted for more than 50%. In that large international study, most of the HPV-39 and HPV-59 types detected were from women in Central and South America; although we found HPV-39 in 4% of our cohort, HPV-59 was not detected in this Mexican population.

It is interesting to compare our results with those reported by Lazcano-Ponce et al3 for a population-based study in Cuernavaca, Mexico (approximately 50 miles from Mexico City). In the Cuernavaca study, 1300 women with normal cervical cytology were tested and typed for HPV DNA. Twenty-four types were detected, and as expected in such a population, 17% of all women testing positive were infected with low-risk types, either alone or in combination with other low-risk types. Not unexpectedly, in our cohort of carcinomas, we found only 1 low- risk type, HPV-84, which was detected in 2 tumors. Nevertheless, the most commonly detected types in the Cuernavaca study (either alone or in combination with other types) were those considered to belong to the category of high-risk HPVs: HPV-16 was found in 13.7% of all positive women, HPV-53 in 13.7%, HPV-31 in 11.3%, and HPV-18 in 8.3%. Similar to our data, in which dual infections were detected in 9 (21%) of 43 of the typed tumors, 20% of the HPV DNA–positive women in the Cuernavaca study had multiple infections. However, unlike our study, in which only single or dual types were detected, the Cuernavaca study had many women with more than 2 HPV types. These discrepancies may reflect the differences in the 2 Mexican cohorts, that is, normal tissue versus tumor, as suggested by Kleter et al.14 In that study, the prevalence of multiple HPV infections differed between women with normal cytology, mild dyskaryosis (dysplasia), and carcinoma. We assume that tumor, as opposed to general sampling by swab or lavage, would tend to be more clonal in nature.

The types of HPV DNA detected in our dual infections are of interest. Thomas et al,22 in a study of university women from Seattle, Wash, investigated the rates of acquiring different HPV types and found no evidence for clustering or cross-protection. The most likely cross-protective effects would be on the basis of phylogenic similarities between different HPV types. Like those data, we saw no evidence of cross-protection in our cohort. Our dual infections were often those of 2 viruses belonging to the same clade, for example, HPV-16, HPV-31, and HPV- 33 all belong to the A9 clade.5 

In conclusion, we examined a group of cervical tumors with a large percentage of unusual or rare histologic types that were archived at the Institute of Cancer in Mexico City. A predominance of anaplastic and adenoid basal carcinoma types was found among these rare types. The HPV types detected in these cervical carcinomas from Mexican women showed distribution patterns similar to carcinomas of the cervix in the United States, with a predominance of HPV-16, even among the rare types of these neoplasias.