Spontaneous Regression of Grade 3 Vulvar Intraepithelial Neoplasia Associated with Human Papillomavirus-16–Specific CD4⁺ and CD8⁺ T-Cell Responses

INTRODUCTION

Cellular immune responses play a critical role in human papillomavirus (HPV) infections by controlling or eliminating the virus, and the incidence of HPV-induced diseases is increased in T-cell immunodeficient individuals, such as HIV-infected or transplanted patients (1, 2, 3) . Numerous HPV are responsible for uterine cervical cancers, most frequently worldwide HPV-16 (4) . In infected patients with high-grade cervical intraepithelial neoplasia (CIN 2/3) or invasive cervical carcinoma, blood cytotoxic T lymphocytes (CTL) directed against HPV-16 are barely detectable (5, 6, 7, 8, 9, 10) , and no in situ functional studies were completed although in situ cervical cytobrush allows infiltrating lymphocytes to be obtained (11) . The impairment of HPV-16–specific CTL responses could relate to a down-regulation of MHC class I molecules on HPV-16–infected cells (12) or to T-cell tolerance to HPV proteins (13) . CD4-proliferative responses to HPV-16 seem to correlate with the stage of infection. Indeed, high frequency specific interleukin-2–producing CD4+ lymphocytes have been observed in asymptomatic HPV-16–infected women whereas they decrease during disease progression toward high-grade CIN or invasive cancer (14, 15, 16) .

To better understand the role of CD4 and CD8 lymphocytes in the defense against HPV-16 infection, we studied blood cellular immune responses in grade 3 vulvar intraepithelial neoplasia (VIN3), also known as bowenoid papulosis (BP). The lesions commonly resemble persistent warts that are multiple and disseminated on the vulva and/or the perianal skin of young female adults. At the pathologic level, there is a squamous cell carcinoma of the whole epithelium (VIN3).

In the present study of 6 patients, we evaluated blood T-cell responses against HPV-16 E6 and E7 proteins. Eighteen overlapping peptides (15-mer to 24-mer) spanning the entire length of both proteins were synthesized with consideration for putative anchor motifs for several HLA class I molecules. Simultaneous vulvar biopsies were submitted to pathologic and immunohistochemical analysis. The very few patients under study prevents any statistical analysis, but the differences between them and their correlation with clinical outcome are so striking that we believe them to be meaningful.

MATERIALS AND METHODS

Patients.

Six BP patients entered the study. HLA class I and class II antigens were determined. These women, aged 24 to 50 years, mean 41 ± 9 years (Table 1) ⇓ had skin or mucosal lesions suggestive of VIN3 and were followed up for at least 6 months. Diagnosis was confirmed by standard pathologic analysis. HPV-16 was isolated from the lesions of all patients. BP first symptoms had appeared from 6 to 156 months before inclusion in the study. At inclusion, five patients had suffered from recurrent lesions for >6 months. Three of them (patients 1–3) had numerous relapses despite multiple destructive treatments (cryotherapy, electrocoagulation, or laser vaporization) or local chemotherapy (5-fluorouracil). At the time of entry in the study, these patients had not received any treatment for 2 months. Treatment with imiquimod cream was ineffective in patient 4. Patient 5 was not treated before inclusion. In patients 1 to 5, VIN3 lesions persisted for at least 6 months after evaluation.

Patient 6 entered the study 6 months after the onset of disease. Electrocoagulation of <50% of the VIN3 lesions was done at that time. Two months later, all vulvar lesions had disappeared, and the patient was clinically disease-free for 3 years.

Blood Samples.

In accordance with the Ethics Committee of Cochin hospital, 150-mL blood samples were collected the day of the entry in the study in every patient after obtaining their informed consent. Another sample was obtained from patient 6, 6 months later to generate short-term effector cell lines.

Peripheral blood mononuclear cells (PBMCs) were isolated by centrifugation through lymphocyte separation medium (Pharmacia, Uppsala, Sweden) and either used immediately or frozen with 10% DMSO and stored at −180°C in liquid nitrogen.

HPV-16 Typing.

HPV-16 typing was done by PCR with DNA extracted from keratinocytes followed by restriction mapping of the amplified products. Multiplex PCR was done with specific E6 HPV-16 and HPV-18 primers as described previously (17) . HeLa and SiHa cell lines were used as negative and positive controls, respectively. After 40 cycles of amplification, products were analyzed on 5% polyacrylamide gels. When a HPV DNA band was detected, the amplified product was digested with restriction enzymes. The appropriate restriction pattern of amplified products, together with its size, confers virtually 100% specificity on the PCR reaction.

Synthetic Peptides.

Eighteen overlapping peptides (15-mer to 24-mer) spanning the entire length of the E6 and E7 proteins (Table 2) ⇓ were synthesized by Neosystem (Strasbourg, France).

Ex Vivo ELISPOT Assay for Single Cell IFN-γ Release.

ELISpot-IFNγ assays were done as described previously (18) . Briefly, nitrocellulose plates (Multi-Screen HA, Millipore, Bedford, MA) were coated overnight at +4°C with 0.1 μg of mouse antihuman IFN-γ monoclonal antibody (mAb; Genzyme, Russelheim, Germany). Plates were washed with PBS and blocked with RPMI 1640-glutamax medium (Life Technologies, Inc., Paisley, Scotland) supplemented with penicillin (100 UI/mL), streptomycin (100 μg/mL), nonessential amino acids (1%), HEPES buffer (10 mmol/L, Life Technologies, Inc.), sodium pyruvate (1 mmol/L; ICN, Costa Mesa, CA), and 10% heat-inactivated human AB blood group serum (complete medium) for 2 hours at 37°C. PBMCs (4 × 10⁵ cells/well in 100 μL) were incubated in duplicate with 5 μmol/L of each peptides in complete medium with 50 UI/mL interleukin-2 (Boehringer, Mannheim, Germany) for 48 hours. Plates were then washed, and 100 μL of polyclonal rabbit antihuman IFN-γ antibodies (Genzyme) at 1:250 dilution were added. After overnight incubation at +4°C, plates were washed and 100 μL of polyclonal biotin-conjugated goat antirabbit IgG antibodies (Boeringher) diluted 1:500 were added and incubated for 2 hours at 37°C. The plates were washed and incubated with alkaline phosphatase-labeled extravidin (diluted 1:5,000, Sigma-Aldrich Chimie SARL, Lyon, France) for 1 hour. Chromogenic alkaline phosphatase substrate (Bio-Rad Laboratories, Hercules, CA) was added to the wells to develop spots. Blue spots were counted with an automatic microscope (Zeiss Apparatus; Carl Zeiss, Göttingen, Germany). Negative controls were PBMCs incubated in complete medium alone. Positive controls were obtained by activating PBMCs with 50 ng/mL phorbol myristate acetate and 500 ng/mL ionomycine (2,000 cells/well, Sigma-Aldrich Chimie SARL). Only large spots with fuzzy borders were scored as IFN-γ-spot forming cells (SFC). Responses were considered significant when the mean number of SFC by 10⁶ cells in 2 experimental wells was superior to the highest of either the mean number of SFC in the negative control (PBMCs alone) plus 3 SDs or the number of SFC in the negative control (PBMCs alone) plus 25 SFC/10⁶ cells.

Generation of Short-Term Effector Cell Lines.

PBMCs (2.5 × 10⁶/mL) obtained from patient 6 after 6 months of follow-up were cultured in 24-well plates (Costar, Cambridge, MA) in complete medium, with pooled E6/4, E7/2, and E7/3 peptides (5 μg/mL) for 12 days. Interleukin-2 (20 IU/mL) was added on days 3 and 7. On day 12, effector cells were washed three times, resuspended in complete medium and used in ELISpot-IFNγ assays as above.

Characterization of IFN-γ Secreting Cells by Specific T-Cell Depletion Assay.

Short-term effector cell lines were incubated on ice for 30 minutes in 500 μL of complete medium with BioMag beads (PerSeptive Diagnostics, Cambridge, United Kingdom) coated with a goat antimouse IgG antibody (50 μL of coated beads/10⁶ cells), and either with a anti-CD8 mAb mixture [a mixture of three mAb (1 μg of each), namely DK25 from Dako, Glostrup, Denmark; Leu 2a from Becton Dickinson, Mountain View, CA; and OKT8 from Ortho Diagnostic Systems, Raritan, NJ) or with two anti-CD4 mAb (1 μg of each; M310 from Dako and OKT4 from Ortho Diagnostic Systems). Conjugate-coated cells were then magnetically removed, and the remaining cells were tested in the ELISpot assay as described above. Positive selection experiments were done also, but they are hardly interpretable because of the activation by the selecting mAb themselves.

T-Cell Proliferation Assay.

PBMCs (2 × 10⁵/200 μL) were cultured in 96-well round-bottomed microtiter plates in complete medium with individual antigenic peptides in triplicate. After 5 days of culture, 1 μCi of [³H]TdR (NEN, Paris, France) was added to each well for 18 hours. Cells were harvested with an automatic cell harvester (Skatron, Sterling, VA) and [³H]thymidine incorporation was quantified by scintillation counting. Proliferative responses with a stimulation index (SI = cpm in presence of antigen/cpm in control media) above 2.5 were scored as positive.

Immunohistochemical Analysis.

Vulvar biopsies done for diagnostic purposes were fixed in formaldehyde. Five micrometer-thick paraffin-embedded fixed sections were stained with H&E for standard pathologic analysis and stained with anti-CD3 (PC3/188A clone, Dako) and anti-CD8 (C8/144B, Dako) mAb after antigen retrieval by heating in a microwave oven in 10 mmol/L citric acid buffer (pH6) and endogenous peroxidase blocking. For CD4 staining with the mAb 4B12 (Novocastra, Newcastle upon Tyne, United Kingdom), antigen retrieval was done with TRIS/EDTA (pH8). The mAb were detected by a biotinylated goat antimouse IgG1 antiserum and streptavidine-biotin complex (Dako) and 3,3-diaminobenzidine tetrahydrochloride as a chromogen.

RESULTS AND DISCUSSION

T-cell responses were tested against E6 and E7 long peptides covering the whole proteins by ex vivo ELISpot-IFNγ. The five patients (1, 2, 3, 4, 5) who presented with chronic recurrent lesions for >6 months had no T-cell responses detectable by the ELISpot-IFNγ assay (Fig. 1) ⇓ . The latter method revealed that the blood of patient 6 at the time of entry in the study contained numerous cells recognizing three HPV-16 long peptides: E6/4 [amino acids (aa) 45–68], E7/2 (aa 7–27), and E7/3 (aa 21–40; mean SFC/10⁶ PBMCs: 270, 65, and 430, respectively; Fig. 2 ⇓ ). Such a response has not been reported previously during HPV-16 infection clearance and is similar to that observed in viral systems such as HIV or EBV in which the viruses are highly replicative or chronically reactivated (19) and in which CD8 lymphocytes play a key role in controlling viral replication. Recently, in two patients with VIN3, ex vivo frequencies of specific anti-E6 or E7 peptides CD8+ T lymphocytes varied between 21 and 1360 SFC/10⁶ CD4-depleted T lymphocytes (20) . However, no clinical correlation was reported in the latter study.

In the second study of the blood lymphocytes of patient 6 done 6 months later, ex vivo ELISpot-IFNγ assay revealed T-cell responses against E6/4 peptide (65 specific spots/10⁶ PBMCs) whereas responses against E7/2 and E7/3 peptides were no longer detectable (data not shown). Hence, a decrease in the number of blood anti-HPV-16 effector cells followed the clinical disappearance of the lesions. Short-term cell lines obtained after stimulation of the PBMCs of patient 6 with the pool of the peptides E6/4, E7/2, and E7/3 and depleted in either CD4 or CD8 lymphocytes were studied by ELISpot-IFNγ assays. Peptides E6/4 (aa 45–68) and E7/2 (aa 7–27) were recognized by the CD4-depleted lymphocyte population (Fig. 3A) ⇓ and not by the CD8-depleted population (Fig. 3B) ⇓ . Peptide E6/4 (aa 45–68) includes peptide E6 52–60 (IVYRDGNPY), which contains binding motifs for HLA-A2 (21) and HLA-B7 molecules, which are present in patient 6. Moreover, region E7/2 (aa 7–27) contains the epitope E7 11–20 that was described previously as a target of CTL in association with the HLA-A2 molecule (21 , 22) . Effector lymphocytes recognizing E7/3 (aa 21–40) peptide were no longer observed in these short T-cell lines and, therefore, could no be characterized. Their absence could be related to an exhaustion of anti-E7/3 effector T cells in vivo or in vitro.

In the proliferative assays done at entry in the study, T lymphocytes from three of the six patients (2 , 5 , and 6) proliferated when stimulated by one or two HPV-16 E6 long peptides (Fig. 4A-C) ⇓ . The peptide E6/4 (aa 45–68) induced the strongest proliferative response in all three patients, with proliferation indexes ranging from 3 to 10. The peptide E6/2 (aa 14–34) was also recognized by T lymphocytes from patients 2 and 5 with an average stimulation index of 6.

These results are consistent with CD4+ T-cell responses because E6 protein peptides are known inducers of proliferative responses. Regions E6 1–31, 22–51, and 24–45 (included in peptide E6/2) are immunogenic for CD4⁺ T cells in CIN or sexually active healthy women (23 , 24) . Region E6 42–71 including E6/4 peptide (aa 45–68) is also as a target of proliferative responses in patients with CIN (23) . However, in view of results obtained in patient 6 with short-term cell lines, CD8+ T cells may also be involved in the proliferative res-ponses. The epitopes E6/2 and E6/4 hence could be strongly recognized by CD4+ and/or CD8+ T lymphocytes and could be particularly relevant in the design of a peptide vaccine.

In every case, immunohistochemical study of vulvar biopsies showed dermal T-cell infiltrates containing both CD4+ and CD8+ lymphocytes with a marked CD4+ infiltrate in patient 6 (Fig. 5A-C) ⇓ . Intraepidermal lymphocytes were observed in patients 4 and 6. Only in patient 6, the infiltrate was made up of both CD4⁺ and C8⁺ T cells and encompassed the entire thickness of the epidermis (Fig. 5B and C) ⇓ , together with apoptotic keratinocytes, whereas in patient 4 intraepidermal lymphocytes were all CD8-positive (Fig. 5G-I) ⇓ . No intraepidermal infiltrate was observed in the biopsied tissues of the four other patients (as illustrated for patient 1 in Fig. 4D-F ⇓ ). Our observations in patient 6 are similar to those reported in infiltrates observed in regressive HPV lesions as plane cutaneous warts (25) or regressive condylomas (26) , which are marked by a strong CD4+ and CD8+ infiltrate around the HPV-infected keratinocytes. On the contrary, in chronic and nonregressive CIN3, lymphocyte infiltrates of the epidermis contain mainly CD8⁺ lymphocytes and no CD4⁺ cells (27 , 28) . It is likely that CD8⁺ lymphocytes play a major role in the defense against HPV infections by killing infected keratinocytes. However, CD4⁺ lymphocytes are required for an optimal induction of CD8⁺ effectors (29, 30, 31) and for activation of tumor-specific memory CD8⁺ T cells (31) . This could explain why patient 4, whose epidermal T lymphocyte infiltrate contained no detectable CD4⁺ T cells, was unable to eliminate the HPV-induced lesions. Clearance of condylomas or CIN has been associated with a Th1 cytokine profile (32) , positive intradermal skin tests against HPV-16 E7 peptides, and presence of CD4+ T lymphocytes (33) .

In conclusion, the differences in the T-cell responses in patient 6 (in whom clinical lesions disappeared within a few months) and in the nonregressive patients are strong enough to tentatively give a clue for a better understanding of HPV control by the cell-mediated immune responses. Clinical regression might have occurred before development of effector T cells inefficiency. Indeed, patient 6 was the only one to show a strong blood T-cell response detected by ex vivo ELISpot-IFNγ and an epidermal and dermal CD4+ and CD8+ T-cell infiltrate before a complete clinical regression of BP lesions. The presence of many blood anti-HPV-16–specific T cells might result from a very high in situ response. Such a study of blood cellular immune responses together with the analysis of vulvar biopsies obtained simultaneously and correlated to clinical outcome was not reported previously. In a recent anti-HPV vaccine trial conducted by Davidson et al. (34) , VIN3 lesions regressed completely in a patient after vaccination. Interestingly, immunostaining of vulvar biopsy before vaccine showed a marked CD4+ and CD8+ T lymphocyte infiltrate. One may wonder whether the regression of this patient’s lesions could be related to spontaneous regression. Therefore, the observation of a CD4+ and CD8+ infiltrate within dermal and epidermal sheets in the biopsy and the visualization of very strong blood anti-HPV T-cell responses in patient with VIN3 could be predictive of spontaneous clinical outcome. In addition, it may also be thought that high-level blood CD4⁺ and CD8⁺ lymphocytes after therapeutic vaccination could allow clearance of HPV-16 lesions in BP, assuming that anti-HPV vaccine-induced T effector cells home in the HPV cutaneous and mucosal lesions.