Limited evidence of human papillomavirus on breast tissue using molecular in situ methods



This article is corrected by:

  1. Errata: Erratum: Limited evidence of human papillomavirus on breast tissue using molecular in situ methods Volume 118, Issue 9, 2561, Article first published online: 6 September 2011



Human papillomavirus (HPV) has been proposed as an etiologic agent of breast cancer based on numerous reports of high-risk (oncogenic) HPV types in malignant breast tissues. However, most of those studies used standard and nested solution polymerase chain reaction (PCR) techniques, both of which are disadvantaged by vulnerability to laboratory contamination from positive control DNA and the inability to localize the signal to a specific cell type. To overcome these drawbacks, the authors of this report explored the use of in situ molecular methods of viral detection to reassess the frequency of HPV in malignant breast tissue.


In situ hybridization (ISH) was used with probes that were specific for the capsid region of 12 oncogenic HPV types, and in situ PCR (IS-PCR) was used with primers that were specific for the capsid region of HPV-16, which is the most common oncogenic HPV type. These methods were resistant to molecular contamination and allowed identification of the positive cell type. The specimens examined were malignant tissues from patients with 70 breast cancer patients at The University of Texas M. D. Anderson Cancer Center in Houston, Texas.


HPV was observed in 4 of 70 specimens (5.7%) using ISH and in 2 of 70 specimens (2.9%) of specimens using IS-PCR. Concordance between the 2 methods was high for negative specimens; both methods yielded negative results in 66 of 70 specimens (94.3%). However, there was no concordance for the few positive specimens, probably because of differences in sensitivity and the targeted HPV types.


Oncogenic (high-risk) HPV types were present in malignant breast epithelium very infrequently and, thus, may be causative agents of only a relatively small proportion of all breast cancers. Cancer 2012;. © 2011 American Cancer Society.


Human papillomavirus (HPV) has been investigated over the past 18 years for its possible association with breast cancer. The majority of these studies (17 of 22)1–17 identified high-risk (oncogenic) HPV types in malignant breast tissues, and the studies that also tested nonmalignant tissues identified HPV more frequently in malignant tissues than in nonmalignant tissues.1–3, 7, 14, 16 This stimulated much interest in the possibility that oncogenic HPV types may play a role in the development of breast cancer.

Most of the studies, however, used standard solution polymerase chain reaction (PCR) and nested PCR, both of which require tissue digestion before DNA extraction. This has 2 disadvantages: a high risk of molecular contamination, especially by positive control DNA present in the same laboratory, and the inability to confirm that a positive reaction is in mammary epithelial cells, the cell type usually affected in breast cancer. There are many other cell types present in digested breast tissues, including leukocytes, adipocytes, fibroblasts, and endothelial cells. In contrast, in situ molecular methods avoid these problems.18 They use formalin-fixed tissue sections on glass slides and observe the end result with brightfield or fluorescent microscopy. In situ hybridization (ISH) results in the binding of labeled DNA probes of specific sequences with homologous sequences in tissue sections and detection of the label within cells. In situ-PCR (IS-PCR) uses a labeled nucleotide incorporated directly into DNA of the target gene, which is amplified during a PCR reaction with the intact tissue section. Both in situ molecular methods allow identification of the cell type that is positive. In addition, neither method is subject to molecular contamination. If a positive molecule from another source contaminated the hybridization mix, then it would not be detected, because it was not labeled. An unwanted molecule in the IS-PCR reaction mix might be amplified and labeled in the solution; however, it would be eliminated in the rinse steps, whereas the target molecules amplified in cells would remain anchored within the tissue. In situ molecular methods were used in only 4 studies that investigated the presence of HPV in breast tissues.4, 5, 7, 8 Three of those studies compared solution PCR and in situ methods for the same set of specimens and identified oncogenic HPV in mammary epithelium of only 4 of 34 (12%) of the collective total, as opposed to 34 of 110 (22%) using standard or nested solution PCR. This raises the possibility that many of the positive results obtained by solution PCR may have been caused by contamination or by the presence of HPV in nonmammary epithelial cell types.

Because of their sensitivity, ability to localize the signal, and resistance to contamination, in situ molecular methods have the potential to establish more accurately the frequency of HPV in the mammary epithelium of breast cancer tissue. Also, because of increasing technologic improvements, these methods have potential as cost-effective tools to detect early markers of breast cancer and need to be studied more thoroughly. We report here the results of using ISH and IS-PCR to search for oncogenic HPV types in a series of tissues obtained from 70 patients with breast cancer at a major cancer center.


The conduct of this study was approved by the institutional review boards of the M. D. Anderson Cancer Center, the University of California San Francisco, and the University of California, Berkeley. All patients provided written informed consent to participate.

Study Population and Material Collection

Breast tissues from 70 women who underwent surgery for breast cancer were obtained between 2008 and 2009 at The University of Texas M. D. Anderson Cancer Center. Patients with in situ and invasive tumors who did or did not receive neoadjuvant chemotherapy were eligible for the study. Six of the 70 patients enrolled for the study received preoperative chemotherapy. Thirty-nine patients underwent mastectomy, and 31 underwent segmental mastectomy. All tissue specimens were processed within 24 hours upon receipt by the study pathologist to minimize any DNA degradation. Tissues were fixed in 10% buffered neutral formalin, processed routinely, and embedded in paraffin. Five-micrometer-thick tissue sections were used for hematoxylin and eosin staining for conventional histopathologic examination and ISH and IS-PCR evaluation.

In Situ Hybridization

ISH used the INFORM HPV III family 16 probe(B) (Ventana Medical Systems Inc., Oro Valley, Ariz), which contains probes for HPV genotypes 16 (HPV-16), HPV-18, HPV-31, HPV-33, HPV-35, HPV-39, HPV-45, HPV-51, HPV-52, HPV-56, HPV-58, and HPV-66. The detection of probes used the ISH iVIEW Blue plus kit, consisting of a primary rabbit antidinitrophenyl (anti-DNP) antibody and an indirect biotin-streptavidin system. Tissue sections were deparaffinized, warmed to 90°C, and treated with 1 drop of protease for 3 to 4 minutes. Subsequently the INFORM HPV-16 probes were applied, and the slides were incubated for 2 hours at room temperature for the hybridization step. Slides were then treated with rabbit anti-DNP primary antibody, which detects the DNP-labeled probes bound to the target sequence. At the end of each incubation step, the sections were rinsed with reaction buffer in a Ventana automated slide stainer to stop the reaction and remove unbound material that would hinder the desired reaction in subsequent steps. Finally, the slides were treated with the amplification reagent (mouse anti-rabbit antibody) followed by the binding of a biotinylated secondary antibody (goat antimouse immunoglobulin G). Streptavidin-conjugated alkaline phosphatase was then used as a chromogenic enzyme and generated a visible blue signal if the probe hybridized to the target sequence. Slides were then counterstained and coverslipped. Positive signals were recognized as blue dots in the nucleus or cytoplasm of the cells. Every run included positive control slides (known HPV-16/HPV-18–positive cervical cancer tissues) with a visible reaction product in the cells and negative controls, which were adjacent breast tissue specimens processed through identical procedures except that probes were omitted from the hybridization solution.

In Situ-Polymerase Chain Reaction

The IS-PCR procedure tested only for HPV-16. The positive control was the established human cell line CaSki (catalog no. CRL-1550; American Type Culture Collection, Rockville, Md),19 which contains approximately 600 copies of HPV-16 per cell. It was obtained directly from the American Type Culture Collection and was used within 2 months after receipt. Its authenticity for our purposes was that it tested positive for HPV-16 and, thus, would be a reliable positive control. It was maintained in Dulbecco modified Eagle medium supplemented with 10% fetal bovine serum, 10 μg/mL insulin, 6.5 μg/mL polymixin B sulfate, 100 μg/mL streptomycin sulfate, 253.5 U/mL penicillin, and 2.5 μg/mL amphotericin B.

The protocol for IS-PCR was based on the method described by Nuovo.20 To prepare the CaSki-positive control cells for IS-PCR assays, trypsinized monolayers were removed from the plastic flask with a standard saline-trypsin-versene solution, rinsed once with Dulbecco phosphate-buffered saline, and centrifuged at low speed (×500g). A thick suspension of the pelleted cells was smeared onto high-adherence glass microscope slides (catalog no. 12-550-15; Fisher Scientific, Hampton, New Hampshire), air dried, and fixed for 18 hours in 10% buffered neutral formalin. To permeabilize cells, control cells and deparaffinized tissue sections (4 μm thick) on high-adherence microscope slides were digested with 2 mg/mL pepsin for 30 minutes in CaSki cells, and for 60 minutes in tissue sections.

The IS-PCR reaction mix consisted of 4.5 mM MgCl2, 0.4 mM deoxyribonucleotide triphosphates, 1 μM primers (synthesized by Operon, Inc., Huntsville, Ala), 8 μM digoxigenin-11-deoxyuridine triphosphate (DIG), 0.06% bovine serum albumin, and 5 U/μL Taq polymerase. Primer sequences for HPV-16 detection were from the L1 capsid region and were amplified a 105-base pair (bp) product21 with positions within the genomes indicated by bp numbering: 5′ to 3′ (bp 1288-1311; TTTGGTCTACAACCTCCCCCAGGA) and 3′ to 5′(bp 1392-1369; TTCTTTAGGTGCTGGAGGTG TATG). Specimens were surrounded with gaskets (Easiseal; Hybaid, Ashford, United Kingdom), 60 μL of PCR reaction mix were placed into the area surrounded by the gasket, and a plastic film (Hybaid) was sealed over the gasket to prevent evaporation. We used Amplitaq gold polymerase (catalog no. 4311806; Applied Biosystems, Carlsbad, Calif), which is engineered to activate only at temperatures >60°C. This reduces the nonspecific DNA repair and other mispriming activity of Taq polymerase that may occur at lower temperatures and can result in a false-positive reaction in the nucleus.20 Slides were placed into the IS-PCR machine (Hybaid) for 1 cycle at 93°C for 10 minutes, 92°C for 2 minutes, and 57°C for 1.5 minutes; for 30 cycles at 92°C for 30 seconds, 57°C for 1.5 minutes, and 69°C for 2 minutes; and for 1 cycle at 69°C for 10 minutes. The location of DIG incorporated into PCR amplicons was detected by an immunoperoxidase assay using anti-DIG antibodies (1:100 dilution) reacted for 1 hour at room temperature. The chromogen was diaminobenzidine, and no counterstain was used initially so that visualization of the signal would not be obscured. A semiquantitative judgment of color density of the cellular reactions was the outcome measurement using a scale from 1 to 4+, with 1+ indicating light tan, 2+ indicating medium tan, 3+ indicating dark tan, and 4+ indicating almost black. Only ratings ≥2+ were counted as positive in this study. Controls, which were run simultaneously and under identical conditions as the experimental samples, included 1) a smear of the CaSki cell line reacted with complete PCR mix (positive control), and 2) a smear of CaSki cells and an adjacent serial section of each tissue reacted with the PCR mix minus the primers (negative controls).

Pathology interpretation

All specimens were evaluated by our team pathologist, S. Krishnamurthy, as part of the clinical diagnosis of the patients. The tumors were graded by using the combined Nottingham histologic grading system. All HPV-positive specimens were reviewed by Dr. Krishnamurthy to evaluate which cell types (normal, premalignant, and/or malignant) were positive for HPV.21 Cells were classified as premalignant according to the consensus guidelines of the Cancer Committee of the College of American Pathologists based on the relative risk of developing breast cancer for women diagnosed with particular nonmalignant breast changes.22

Statistical Considerations

All statistical analyses were performed in SAS version 9.2 (SAS Institute Inc., Cary NC). We used the following procedure (PROC) terms for all analyses: 1) PROC MEANS to calculate the mean, median, standard error, and standard deviation for all continuous variables (age, estrogen receptor status, progesterone receptor status, and patient tumor size); 2) PROC FREQ (frequency) with option CHISQ to perform chi-square tests to test whether women who were positive for “any” virus were more likely to have a specific tumor grade after categorizing the tumors as NEWGRADE = 0 (low, low/intermediate), NEWGRADE = 1 (intermediate), or NEWGRADE = 2 (high, intermediate/high); 3) PROC NPAR1WAY (Wilcoxon option) to compare the means of estrogen receptor and progesterone receptor values between women who were positive for “any” virus and women without virus (this test was stratified further by race to check whether racial/ethnic differences may factor into the comparison); and 4) PROC TTEST also was used to compare the means of tumor sizes between women who were positive for “any” virus and women without virus. All comparisons tested the null hypotheses that the populations were the same using a 2-tailed P value < .05 as the level of significance.


Population Demographics

The characteristics of the study population and their malignant breast tumors are summarized in Table 1. The mean patient age was 54 years (range, 32-77 years). The tumors were ductal carcinoma (either invasive or in situ) in 65 patients (92.9%) and invasive lobular carcinoma in 5 patients (7.1%). Thirty-eight percent of tumors were designated high grade, 40% were designated intermediate grade, and 22% were designated low grade (including ductal carcinomas in situ). Nineteen percent of women underwent bilateral mastectomies, and 81% underwent unilateral surgery. Tumor sizes ranged from 0.3 cm to 11.0 cm; and 66% of tumors measured ≤2 cm, 26% measured from 2 cm to 5 cm, 7% measured ≥5 cm, and 1% (1 tumor) had no exact measurement recorded. Forty-eight of 70 women (69%) were Caucasian, 12 of 70 (17%) were Hispanic, 5 of 70 (7%) were African American, and 4 of 70 (6%) were Asian.

Table 1. Patient Ages and Malignant Breast Tumor Characteristics
Patient IDAge, yPrimary Tumor Size, cmCancer TypeCancer GradeBilateralRace/Ethnicity
  1. Abbreviations: DCIS, ductal carcinoma in situ; ID, identification; y, years.

20701.3DuctalIntermediateNoAfrican American
28371.7DuctalHighYesAfrican American
36521.8DuctalHighNoAfrican American
47763.3DuctalHighNoAfrican American
50481.1DCISHighNoAfrican American
66442.2DuctalLowYesNot recorded

HPV Detection

Six of 70 specimens (8.6%) were positive for high-risk HPV types, including 2 specimens (2.9%) that were analyzed with IS-PCR and 4 specimens (5.7%) that were analyzed with ISH. Figure 1 illustrates a specimen that was positive for high-risk HPV according to ISH analysis. Signals were noted in mammary epithelial cells of the malignant tumor and adjacent normal lobules and ducts, as detailed in Table 2. Positive reactions for HPV-16 by IS-PCR were localized to mammary epithelial cells of normal lobules and ducts, as detailed in Table 2 and illustrated in Figure 2.

Figure 1.

A specimen from Patient 44 reacted with primers for human papillomavirus (HPV) in an in situ hybridization reaction using the INFORM HPV probe from Ventana Medical Systems Inc. (Oro Valley, Ariz). (A) Positive signals are observed in the specimen of invasive ductal carcinoma at low-grade (×40) magnification. (B) Positive signals are observed in normal lobules adjacent to the invasive tumor.

Figure 2.

A specimen from Patient 12 reacted with primers for human papillomavirus (HPV) type 16 in an in situ polymerase chain reaction (PCR) analysis. (A) This negative control section was incubated with a PCR reaction mix without primers (original magnification, ×40). (B) The adjacent section was incubated with a PCR reaction mix that contained primers. Note the brown areas, which indicate a positive signal for HPV, in the mammary epithelial cell cytoplasm of normal lobules (original magnification, ×40). (C) This is an enlargement of the boxed area in B (original magnification, ×133).

Table 2. Characteristics of Human Papillomavirus-Positive Breast Tissue Specimens
Patient IDTumor TypeTumor GradeMethod of HPV DetectionPathology of HPV-Positive Cells
  1. Abbreviations: DCIS, ductal carcinoma in situ; HPV, human papillomavirus; IDC, invasive ductal carcinoma; ISH, in situ hybridization; IS-PCR, in situ polymerase chain reaction; M, malignant, N, normal; PM, premalignant.


Four of the 6 women (67%) who had positive HPV results were Caucasian compared with 44 of 63 Caucasian women (70%) who were HPV-negative (race was unknown for 1 HPV-negative woman). Three of 6 HPV-positive women were positive for estrogen receptor (50%) compared with 52 of 64 (81%) HPV-negative women (P = .20).

Conventional HPV status as an incidental finding on Digene Hybrid capture 2 (Qiagen, Valencia, CA) or SurePath (TriPath Imaging, Burlington, NC) results from cervical specimens was available from 12 of 70 patients in the study and were negative for HPV. One of these patients was among the 6 who had positive results determined in the current study by ISH analysis. The HPV status by Digene Hybrid capture 2 in the remaining 5 positive patients was not available.

Both in situ methods concurred for positive results for the control cell line (CaSki cells infected with HPV-16). Overall, the concordance between the 2 methods was very high; both methods obtained negative results in mammary epithelial cells from 64 of 70 specimens (91.4%). However, there was no concordance between the 2 methods regarding which individual specimens were positive.


The frequency of oncogenic HPV types we detected in malignant breast tissues (8.6%) was markedly lower than what was reported in many earlier studies, some of which reported frequencies as high as 74% (range, 7%-74%; mean, 33%; median, 30%). A recent meta-analysis of 9 previous studies using careful statistical tools suggested an overall odds ratio of 3.63 for breast cancer risk in women whose breast tissue tested positive for oncogenic HPV types.23 We contend, however, that contamination may be partially responsible for the higher frequencies that were the basis of that meta-analysis. Most previous investigations used solution PCR techniques, either standard or nested PCR, both of which require initial digestion of tissue and extraction of DNA. This eliminates the possibility of ever knowing which cell type harbored the virus and allows for false-positive results caused by laboratory contamination with DNA from positive controls or circulating HPV virions, DNA, or infected cells from other body sites, such as the uterine cervix.8, 15 Of the 9 studies that were included in the meta-analysis, 8 used solution PCR techniques.23 The molecular methods we used, ISH and IS-PCR, are free of these shortcomings.

Our results indicated a low frequency of oncogenic HPV types in breast cancer specimens, which is what would be expected based on the finding that most HPV-related cancers are squamous in origin and that most breast cancers arise in mammary epithelium, which is not a squamous tissue. None of the specimens had koilocytosis, a classic morphologic change in squamous epithelial cells with viral effect. None of the positive cases were squamous in type, and there was no evidence of squamous metaplasia.

Our use of serial sections from the same tissue block allowed application of the 2 methods to the same specimen. The range of tumor grades and breast cancer types we tested (lobular and ductal, in situ and invasive) helped to ensure detection of a virus that might be associated only with a particular cancer type.

Our sample size of 70 patients was comparatively large compared with the samples in previous studies (range, 11-124 patients; mean, 51.5 patients; median, 50 patients), thus reducing misrepresentations caused by sampling error, which may have occurred in some previous studies with small sample sizes. There is an unlikely possibility that our findings are a result of a lower frequency of high-risk HPV types circulating in the population from which our samples were derived. Donors in previous studies were from Asia, Europe, North and South America, and Australia.1–17 The 3 previous investigations of US populations reported frequencies of 0%,24 10%,4 and 35%.13 If the incidence of cervical cancer caused by oncogenic HPV types can be used as a rough indicator of the extent of circulating oncogenic HPV types in a population, then the possibility that geography explains the frequency variation of breast tissue HPV among studies is unlikely. The 2 studies that obtained the highest frequencies of high-risk HPV in breast tissue (China, 68%; Turkey, 74%) were in countries with a low incidence of cervical cancer.25 Tissues from populations with a high incidence of cervical cancer, eg, Mexico, Brazil, and India,25 reported frequencies of oncogenic HPV in breast tissues of 29%, 25%, and 0%, respectively.2, 3, 26 However, more extensive global studies with standardized methodology would be necessary to prove or disprove this concept.

It is interesting to note that, in all of our HPV-positive malignant specimens, HPV was detected in normal mammary epithelium present in the same tissue section. The E6 and E7 proteins of oncogenic HPV types disable the function of p53 and retinoblastoma (Rb) proteins, respectively,27 and thus disable normal cellular growth-control mechanisms (Rb) and facilitation of DNA repair (p53). HPV-infected cells may gradually accumulate unrepaired mutations, thereby increasing genomic instability and potentially leading to malignancy.27, 28 Therefore, it is reasonable to expect the presence of HPV-related DNA in normal tissues before premalignant changes and malignant transformation. According to the field cancerization theory, most cancers are rare events that develop over decades within a population of carcinogen-initiated cells that are widespread in a tissue.29, 30 This pattern occurs in carcinoma of the uterine cervix, and HPV is a well established cause of this such cancer.28 If breast exposure to oncogenic HPV types occurred through a systemic route or a mechanical route externally through the nipples, then it would be reasonable to assume that both breasts would have equal exposure; thus, the field of HPV cancerization would be bilateral. Only 2 of the 6 specimens that were positive for HPV in the current study were from patients who underwent bilateral mastectomy, and we did not test the contralateral breast.

The Ventana HPV system was developed to detect oncogenic HPV types in the uterine cervix. To our knowledge, this study is the first to apply it to breast tissue. A possible limitation of this method is that there is no amplification; therefore, the Ventana system may not be as sensitive as the PCR methods used in other studies. However, it allows localization of the signal; and it is standardized, cost-effective, and capable of detecting 12 different oncogenic HPV types. Our frequency of 5.6% obtained with ISH was close to that obtained by the 3 previous studies using ISH,4, 5, 8 which obtained 10%, 0%, and 5% frequency, respectively. Only 1 previous study used IS-PCR, which involves amplification, to detect HPV in breast tissue.7 Those investigators reported that 8 of 31 (31%) breast carcinoma specimens were positive for oncogenic HPV. Seven of these harbored HPV-18, and only 1 of 31 (3.2%) harbored HPV-16. This agrees well with the 2.9% frequency we observed for HPV-16 using the same method. Our 2 in situ molecular methods had 91.4% (HPV) concordance for indicating no presence of these viruses in malignant human breast tissues, but they were not concordant for any of the positive specimens. Our ISH testing was able to detect 12 different oncogenic HPV types (HPV-16, HPV-18, HPV-31, HPV-33, HPV-35, HPV-39, HPV-45, HPV-51, HPV-52, HPV-56, HPV-58, and HPV-66), whereas the IS-PCR in our study tested only for oncogenic HPV-16. The most likely explanation for the 2 specimens that were HPV-positive by IS-PCR but not by ISH is that IS-PCR may be more sensitive than ISH because of the PCR amplification step. ISH reportedly detected 50 to 100 copies of HPV in cervical tissue, whereas IS-PCR reportedly detected 1 to 10 copies.18 The most likely explanation for the 4 samples that were positive for HPV by ISH but not by IS-PCR is that they were positive for HPV genotypes other than HPV-16 in the multiprobe Ventana system. In both methods, cross-reactivity with nononcogenic HPV types is precluded by the narrow specificity of the probes and primers for oncogenic HPV types.

In conclusion, the objective of the current study was to contribute toward a more accurate establishment of the frequency of oncogenic HPV types in breast cancer tissues. Our data resulting from 2 different in situ molecular methods used on the same specimens suggest that HPV is not present in mammary epithelium at the higher frequencies reported in many other studies and, thus, may be a causative agent of only a relatively small proportion of all breast cancers. Given this finding, ISH screening of breast tissues for multiple oncogenic HPV types probably would not be cost-effective for diagnosis or identification of women at high risk for developing breast cancer. Nevertheless, the presence of oncogenic HPV in 8.6% of the breast cancer specimens we tested warrants further investigation after obtaining breast tissue from age-matched women with no history of breast cancer, the appropriate normal control to test causal association.


We thank the Dr. Susan Love Research Foundation and the Avon Foundation for their support of this project. We thank Shawn Tang in the M. D. Anderson pathology laboratory. We also thank the breast cancer patients from M. D. Anderson Cancer Center who donated their tissues for research.


This work was funded by the Avon Foundation (grant 07-2007-073) and by a pilot grant from the Dr. Susan Love Research Foundation.


The authors made no disclosures.