Background: Infection with human papillomavirus (HPV) type 16 (HPV16) is a major cause of high-grade cervical intraepithelial neoplasia (CIN). Experiments were planned to evaluate the role of cell-mediated immunity (e.g., lymphocyte proliferation) against HPV in the natural history of HPV-associated neoplasia and to identify antigenic sequences of the HPV16 proteins E6 and E7 against which an immune response may confer protection. Methods: Fortynine women with abnormal cervical cytology and biopsyconfirmed CIN were followed through one or more clinic visits. Lymphoproliferative responses of peripheral blood mononuclear cells to HPV16 E6 and E7 peptides were assessed in long-term (3-week) cultures. HPV DNA was detected in cervicovaginal lavage by means of polymerase chain reaction and Southern blotting. Disease status was determined by cervical cytologic examination and colposcopy. Reported P values are two-sided. Results: Subjects with positive lymphoproliferative responses to E6 and/or E7 peptides were more likely to be HPV negative at the same clinic visit than were nonresponders ( P = .039). Subjects who were negative for HPV and those with a low viral load were more likely to be responders than were those with a high viral load ( P for trend = .037). Responses to N-terminal E6 peptide 369 were associated with absence of HPV infection at the same clinic visit ( P = .015). Subjects with positive responses to E6 or E7 peptides at one clinic visit were 4.4 times more likely to be HPV negative at the next visit than were nonresponders ( P = .142). Responses to E6 peptide 369 and/or E7 Cterminal peptide 109 were associated with an absence of HPV infection ( P = .02 for both) and an absence of CIN ( P = .04 and .02, respectively) at the next visit. Methods: Lymphoproliferative responses to specific HPV16 E6 and E7 peptides appear to be associated with the clearance of HPV infection and the regression of CIN.
Human papillomavirus (HPV) infection has been shown to be the major etiologic factor in the development of cervical intraepithelial neoplasia (CIN), the presumed precursor of carcinoma of the uterine cervix ( 1 , 2 ) . Most CIN lesions in normal women regress spontaneously, yet the disease in a subset of women undergoes progression to invasive cancer. Possible risk factors, among others, for the development of high-grade CIN or cancer include persistent HPV infection, high viral load, increased age ( 3 ) , infection with other sexually transmitted diseases ( 2 ) , cigarette smoking ( 4 ) , and human leukocyte antigen (HLA) type ( 5 ) . Evidence from epidemiologic studies ( 6 , 7 ) suggests that an intact cell-mediated immune system is required for host control of HPV-associated disease, yet little is known about the role of specific human cellular immune responses to HPV in the natural history of HPV-induced disease. Immunologic studies in murine model systems ( 8 – 10 ) have suggested that the HPV type 16 (HPV16) E6- and E7-transforming proteins are tumor rejection antigens, yet a direct experimental proof in humans is lacking.
Serologic immune responses to HPV16 proteins E6 and E7 have been described in women with CIN and cervical cancer ( 11 – 14 ) , but their role in disease outcome is unclear. Only a few studies of T-cell responses to E6 or E7 proteins in humans with well-characterized genital HPV-associated disease have been reported ( 15 – 17 ) , so that the association between cellmediated immune responses to specific HPV antigens, clinical status, and HPV status is unclear. Lymphoproliferative responses to specific HPV16 E7 peptides in women with HPVassociated CIN were shown to be associated with infection with HPV16 and closely related types, whereas serologic reactivity to the same peptides did not appear to be HPV type specific ( 18 ) . Another recent study ( 19 ) has shown that cell-mediated immune responses to E7 peptides are associated with HPV type-specific infection and that cell-mediated immune reactivity may be associated with persistent HPV infection. It has been shown recently that patients with cervical cancer have deficient cellmediated immune responses to both L1 and E7 peptides ( 20 ) and that induction of interleukin 2 (IL-2) production after stimulation with E6 and E7 peptides is deficient in women with highgrade CIN or cancer ( 21 ) , suggesting that development of cancer is associated with failure of protective cell-mediated immune responses to HPV.
We evaluated lymphoproliferative responses to HPV16 E6 and E7 peptides in women with HPV-associated CIN and examined the association between immune responses to specific peptides, prevalence of HPV infection, and natural history of associated dysplasia.
Subjects and Methods
Study Population and Study Design
We studied 49 nonpregnant women with recent histories of abnormal cervical cytology and biopsy-confirmed CIN seen in the colposcopy clinic at Jacobi Medical Center, Bronx, NY. At each study visit, colposcopic examination was performed, and cervicovaginal lavage for HPV detection and heparinized venous blood samples for cell-mediated immunity assay were obtained. The 49 subjects contributed a total of 80 observations, which constituted the cross-sectional analysis in which the association was examined between HPV infection and cell-mediated immune reactivity to HPV peptides, assessed from blood and cervicovaginal lavage samples obtained during the same clinic visit (time t ). Twenty subjects were studied at multiple time points and had repeated HPV and cell-mediated immunity measurements.
Among the 49 subjects, 30 women with grade II CIN (moderate dysplasia) ( 22 ) were participants in a randomized, double-blinded, placebo-controlled clinical trial evaluating the efficacy of β-carotene for causing regression of CIN and were followed at 3- to 5-month intervals, for a total of 15 months ( 3 ) . Blood samples for cell-mediated immunity assay were obtained at one or more clinic visits from each of these women, and these study subjects contributed a total of 63 cell-mediated immunity observations. Eighteen subjects contributed multiple cell-mediated immunity measurements, and 12 contributed one measurement each. The clinic visit in which blood sample for cell-mediated immunity assay was obtained was designated as time t. Colposcopy, Pap smear, and HPV results from the clinic visits immediately before and after time t (designated as times t − 1 and t + 1, respectively) were obtained from the clinical trial database and merged with the cell-mediated immunity results at time t. These constituted the observations in the longitudinal analysis, where cell-mediated immune reactivity observed at one clinic visit (time t ) was related to HPV and CIN outcomes at the next visit about 3 months later (time t + 1) and where the effect of HPV status at the previous visit (time t −1) on cell-mediated immune response at the next visit (time t) was examined.
CIN status at each visit was determined by both Pap smear and colposcopy. Abnormal colposcopy was based on clinical detection of dysplasia or condyloma, which was in turn based on the presence of aceto-white foci and both punctation and/or mosaicism on the cervix. A Pap smear was considered abnormal if a low- or high-grade squamous intraepithelial lesion or carcinoma was detected by use of the Bethesda classification system ( 23 ) . CIN was considered to be absent if both colposcopy and Pap smear were normal and present if either one was consistent with cervical dysplasia. The protocol was approved by the local institutional review board, and all subjects signed an informed consent prior to enrollment.
HPV Identification and Typing
HPV DNA was identified in cervicovaginal cells obtained by cervicovaginal lavage and typed by polymerase chain reaction (PCR) and Southern blot hybridization with restriction analysis, as previously reported ( 3 , 24 , 25 ) . HPV typing of amplified PCR products was done by dot blot hybridization by use of probes for 39 HPV types, as previously described ( 25 , 26 ) . For Southern blot hybridization, DNA digested by Pst I was subjected to electrophoresis, blotted, and hybridized with a solution containing 32P-labeled HPV types 11, 16, 18, and 51 DNA. Absence of infection was defined as both PCR and Southern blot hybridization negative for HPV DNA. Samples were classified as having high viral load if both Southern blot hybridization and PCR were positive for HPV DNA and as having a low viral load if PCR was positive and Southern blot hybridization was negative, as previously reported ( 3 , 27 ) .
Lymphocyte Proliferation Assay
Heparinized venous blood samples of 10-15 mL were collected from each subject at each clinic visit. Peripheral blood mononuclear cells (PBMCs) were isolated by use of Leucoprep (Becton Dickinson, Franklin Lakes, NJ). Lymphoproliferative responses were tested by use of long-term cultures, as previously reported (18). Since only rare positive responses were seen after standard 6-day lymphoproliferative cultures, 2- to 3-week cultures with weekly restimulation with peptides were established. When PBMCs were processed, aliquots were cryopreserved (0.75 × 10 6 cells per aliquot) for later use as antigen-presenting cells. Remaining PBMCs were cultured at 1 × 10 6 cells/mL (in RPMI-1640 medium with 6% human AB serum) with E6 and E7 peptides or with medium alone for 1 week. Optimal concentrations of peptides used were originally determined by testing different concentrations. The optimal concentration was determined to be 15 mg/mL for the peptides used. Fresh peptides as well as irradiated (2500 rad) aliquots of PBMCs as antigen-presenting cells were added at 7 and 14 days. Recombinant human IL-2 (15 U/mL; Boehringer Mannheim Corp., Indianapolis, IN) was added on day 14. [3H]Thymidine was added to aliquots of the bulk cultures in flat-bottomed, 96-well microtiter plates at 14 and 21 days (10 µL of 0.1 µCi/mL stock was added per well and cultured for 6-8 hours or overnight). Stimulation indices were calculated as counts per minute (cpm) in the presence of antigenic peptides divided by cpm with medium alone. A positive response was defined as a stimulation index of 4.0 or higher. As a control for T-cell reactivity, PBMCs were cultured in microtiter wells at 2 × 105 cells/well with the mitogens concanavalin A and phytohemagglutinin (3-4 days at 0.0005 and 0.001 mg/mL, respectively) and with Candida antigen for 5-6 days (1:100 dilution; Center Laboratories, Hawthorne, NY).
Each sample was tested for lymphoproliferative responses to individual HPV16 E6 and E7 peptides. Peptides were not pooled because one of our goals was to evaluate immune responses to specific epitopes. A sample was considered cell-mediated immunity positive to HPV16 if a lymphoproliferative response to one or more peptides was demonstrated on either day 14 or day 21 or on both days and negative if a lymphoproliferative response to all peptides tested (stimulation index <4.0) was absent on both day 14 and day 21.
HPV16 E6 and E7 Peptides
E7 C-terminal peptides 108 (amino acids 62-80) and 109 (amino acids 72- 97), which were previously shown to be recognized by PBMCs from HPVinfected subjects, were synthesized, as described previously ( 18 ) . E7 peptide sequences used in this study were DSTLRLCVQSTHVDIRTLE and THVDIRTLEDLLMGTLGIVCPICSQKP, respectively. Since in our previous work ( 18 ) we found that cell-mediated immune responses to N-terminal E7 peptides were observed only infrequently in patients with CIN, these peptides were not included in this study, where we concentrated primarily on E6 sequences and C-terminal E7 peptides 108 and 109.
Eight E6 peptides consisting of 30 residues overlapping by 10 amino acids and covering the entire HPV16 E6 open reading frame ( Fig. 1 ) were made by solidphase peptide synthesis, as previously described ( 28 – 30 ) . The full-length peptides were assigned odd numbers (e.g., 369, 371, etc.). To obtain truncated peptides from the C-terminal regions of the larger peptides in order to begin mapping reactive epitopes, we removed a portion of the resin from the reactions after 15 residues were synthesized. These constitute the even-numbered peptides used in this study (i.e., 370, 378, and 380). Structures of all peptide preparations were verified by electrospray ionization mass spectrometry and amino acid analysis.
Since some subjects had multiple observations that tended to be correlated, a generalized linear model with a generalized estimating equation approach for time-dependent, longitudinal data was used for analyses ( 31 , 32 ) . By specifying an exchangeable “working correlation matrix,” this statistical approach took into account the intrasubject correlation and assumed the correlation between any two distinct intrasubject observations to be the same for all individuals in the study.
In the cross-sectional analysis, correlation between HPV infection and cellmediated immune response measured at a given time point (time t ) was examined in the generalized linear model [logit Pr ( CMI t ) = β 0 + β 1 HPV t , where Pr = probability and β 0 and β 1 are regression coefficients]. The dependent variable was dichotomized as positive or negative for cell-mediated immunity to E6 or E7 peptides in general or for cell-mediated immunity to a specific peptide. The independent variable could be the presence or absence of HPV infection, viral load, or HPV type or the presence or absence of CIN. In the longitudinal analysis, only the 30 subjects who participated in the β-carotene clinical trial and who were seen repeatedly were included. To examine if cell-mediated immunity to HPV16 observed in samples obtained at one clinic visit (time t ) had any protective effects on HPV infection and/or CIN status at the next clinic visit (time t + 1), after adjusting for randomization group assignment (β-carotene or placebo, coded below as GROUP), we used the following general linear models, where β 0 , β 1 , and β 2 are regression coefficients:
Regression coefficients (β 1 + β 2 ) were expressed as odds ratios (ORs). P values presented are two-tailed.
Lymphoproliferative Responses to E6 and E7 Peptides
Since our previous study ( 18 ) showed only infrequent cellmediated immune responses to N-terminal E7 peptides and since this study concentrated on mapping human T-cell epitopes in the E6 protein, only two C-terminal E7 peptides were included (i.e., 108 and 109, amino acids 62-80 and 72-97, respectively), and peptides overlapping the entire E6 ( Fig. 1 ) were studied. Among the 49 subjects tested, 33 (67.3%) had cell-mediated immunity to one or more HPV16 E6 or E7 peptides. Of these 49 subjects, 32 (65.3%) had cell-mediated immunity to at least one E6 peptide and 21 (43.8%) of the 48 tested with E7 peptides responded to peptides 108 and/or 109 ( Table 1 ). Sample cell-mediated immunity experiments from two study subjects with CIN II after 21 days of culture with peptides are illustrated in Table 2 . One donor (upper section of Table 2 ) showed no peptide reactivity. The second subject (lower section of Table 2 ) had reactivity to E6 peptides 369 and 381 and to E7 peptide 109 after 21 days of culture, while reactivity to E6 peptides 379 and 383 was negative. Neither of these subjects responded to any peptides after 14 days of culture.
Among the 32 subjects who had positive cell-mediated immunity to E6 peptides, 20 (62.5%) also responded to at least one E7 peptide, whereas 16 (94.1%) of the 17 subjects who failed to respond to the E6 peptides also failed to respond to the E7 peptides ( P <.001). Thus, positive cell-mediated immune responses to E6 and E7 peptides define a specific population of patients with a history of HPV infection and CIN, and patients who are cell-mediated immunity negative define another subset. Fluorescence-activated cell sorter analysis of peptide-stimulated bulk cultures showed that the vast majority of reactive lymphocytes (70%) were CD4 positive (data not shown). The absence of a response to HPV16 peptides was not likely due to generalized immunosuppression, since mitogen responses (concanavalin A and phytohemagglutinin) and standard recall antigens (Candida) were tested and found to be normal in 28 (87.5%) of 32 peptide responders and in 13 (86.7%) of 15 peptide nonresponders for whom these data were available. The 49 subjects contributed 80 observations at different time points and showed reactivity to different E6 and E7 peptides ( Table 1 ). Thirty-five (87.5%) of 40 PBMC samples that were tested for both E7 peptides 108 and 109 showed concordant results; therefore, peptide 108 was not tested with all samples. Among the E6 peptides, none of the samples responded to peptide 375, and no further analysis was done with this sequence. Most of the full-length (odd-numbered) E6 peptides tested (369, 371, 379, 381, and 383) were frequently reactive and were tested in all subjects and samples with sufficient PBMCs.
To begin to map epitopes in frequently reactive peptides, three shorter peptides (i.e., 370, 380, and 382) from the C terminal ends of the longer peptides were tested in a subset of subjects and samples and were added later in the study ( Fig. 1 ). Reactivity to peptides 370 and 380 appeared to be more frequent than to the full-length peptides 369 and 379, respectively, suggesting that at least some reactive epitopes are present in the C-terminal ends of these longer peptides ( Table 1 ). Increased prevalence of positive lymphoproliferative responses to these peptides may be due to testing later in the course of the subjects' HPV infection, when cell-mediated immune responses are more likely to occur. Responses to the full-length peptides 379 and 381 were compared with responses to the complementary Cterminal truncated peptides 380 and 382. Of 14 subjects who failed to respond to peptide 379, 13 (92.8%) also were nonresponders to the truncated peptide 380, and 18 (81.8%) of 22 nonresponders to peptide 381 were also negative with the shorter overlapping peptide 382, as expected. Of six samples responding to peptide 379, only three (50%) responded to the shorter peptide 380. Of 13 samples responding to peptide 381, seven (53.8%) responded to the shorter overlapping C-terminal sequence 382, suggesting that T-cell epitopes are present equally in both C- and N-terminal halves of peptides 381 and 379, respectively. For peptides 369 and 370, six (75%) of eight responders to the longer peptide also responded to the shorter C-terminal sequence, suggesting that a major epitope is likely present in the C-terminal half of peptide 369. An unexpected finding was that, of 27 samples that failed to respond to the longer peptide 369, seven samples (25.9%) did respond to the shorter peptide 370. Of 13 samples exhibiting positive reactivity to peptide 370 (amino acids 16-30), only six (46.2%) responded to the longer peptide 369 (amino acids 1-30), suggesting that sequences present in the N-terminal half of peptide 369 may somehow interfere with processing or presentation of the C-terminal epitope to responding T lymphocytes.
Associations were examined between cell-mediated immunity to HPV16 peptides and HPV infection, characterized by HPV DNA positivity, HPV type, and viral load. When lavage for HPV detection and blood samples for cell-mediated immunity assay were obtained at the same clinic visit, a negative correlation between cell-mediated immune reactivity and HPV infection was detected. Responders to any one or more HPV16 E6 or E7 peptides were significantly more likely to be HPV negative than were nonresponders (OR = 3.1; 95% confidence interval [CI] = 1.1-9.1; P = .039).
Among the 51 cervicovaginal lavage samples that were HPV positive and typed by PCR, 15 (29.4%) had HPV type 16 or closely related types, i.e., HPV types 31 or 33. Cell-mediated immune responses to E6 and E7 peptides were associated with HPV type, in that subjects infected with HPV16 and closely related HPV types 31 and 33 were more likely to be responders than were subjects infected with other HPV types; the difference, however, did not reach statistical significance (OR = 2.3; 95% CI = 0.4-13.8; P 4 .371). There was a linear association between low HPV load and a cell-mediated immune response, suggesting that low viral load is associated more with responder status than is high viral load. Compared with the OR for HPVnegative subjects who were more likely to be responders, the OR for being a responder with a low viral load was 0.5 (95% CI = 0.09-2.5) and 0.3 for being a responder with a high viral load infection (95% CI = 0.1-0.9), respectively ( P for linear trend = .037).
Cell-mediated immune responses of PBMC samples to specific peptides from subjects who were HPV positive at the same clinic visit were compared with responses of samples from HPV-negative subjects ( Table 3 ). Positive cell-mediated immune reactivity to N-terminal E6 peptide 369 was significantly correlated with absence of HPV infection. The OR for being a responder when HPV-negative samples were compared with HPV-positive samples was 3.4 (95% CI = 1.3-9.3; P = .015), suggesting that this sequence contains a “protective” epitope. Several other peptides, including E6 peptides 379, 380, 381, and 383 and E7 peptides 108 and 109, showed similar increases in prevalence of cell-mediated immune reactivity when HPVnegative samples were compared with HPV-positive samples (OR > 2.0), although these increases were not statistically significant owing to the small numbers of samples tested.
Cross-sectional associations between cell-mediated immune responses and presence or absence of CIN as detected by Pap smear and/or colposcopy at the same clinic visit showed an OR for presence of cell-mediated immunity of 2.4 (95% CI = 0.6- 10.2; P = .233) in subjects with absent CIN compared with subjects with CIN who were more likely to be nonresponders, again suggesting an association between cell-mediated immune responses to specific E6 and E7 peptides and absence of disease, although this result was not statistically significant.
Samples obtained from 30 subjects in a β-carotene clinical trial were tested for cell-mediated immunity to HPV16 peptides. Among the 63 samples from these 30 subjects, 56 were obtained after the initial study visit and hence had HPV results from the previous clinic visit (at time t − 1). Forty-one samples were obtained before the last visit of the clinical trial. These samples had HPV results as well as Pap smear and colposcopy from the next visit (at time t + 1). HPV positivity at one clinic visit did not predict cell-mediated immunity at the next visit. The OR for being a responder at the next visit if HPV was positive at the previous visit, adjusted for randomization group (β-carotene versus placebo), was 0.8 (95% CI = 0.2-3.3; P = .755). HPV types 16, 31, or 33 (versus other HPV types) or viral load determined from samples obtained at the subject's previous visit also did not correlate with cell-mediated immunity at the next visit (data not shown).
After we controlled for the randomization group (β-carotene versus placebo), positive cell-mediated immunity to E6 or E7 peptides at one clinic visit (time t ), however, tended to predict the HPV result at the next visit 3 months later (time t + 1). Responders to any one or more E6 or E7 peptides at one visit were four times more likely to be HPV negative at the next clinic visit than were nonresponders to any one or more peptides, who were likely to be HPV positive (adjusted OR = 4.4; 95% CI = 0.6-31.1; P = .142), suggesting a trend, although not statistically significant. Table 4 shows the effects of cell-mediated immune responses to specific E6 and E7 peptides at one visit (time t ) on status of HPV infection at the next visit 3 months later (time t + 1). Compared with nonresponders, positive lymphoproliferative responses to E6 peptide 369 and E7 peptide 109 had the strongest predictive values for subsequent absence of HPV infection and thus appeared to have “protective” activity (adjusted ORs 4 5.0 [95% CI = 1.3-18.8] and 6.7 [95% CI = 1.3-33.2]; P = .018 and .021, respectively).
Thirty-two samples had cell-mediated immunity measurement assessed by use of PBMCs obtained at one clinic visit (time t ) and status of CIN, determined by Pap smear and colposcopy, at the next visit. Responders to one or more peptides at one visit were 3.7 times more likely to have absent CIN (normal Pap smear and colposcopy) at the next visit about 3 months later than were nonresponders (adjusted OR = 3.7; 95% CI = 0.5-27.5; P = .209). Subjects with positive cell-mediated immunity to E6 peptide 369 and E7 peptide 109 at one clinic visit were significantly more likely to have lost their CIN (regression) by the next visit. ORs were 6.5 (95% CI = 1.1-39.1; P = .039) and 6.1 (95% CI = 1.1-35.0; P4 .021), respectively ( Table 5 ).
We present here data on cell-mediated immune responses to overlapping peptides covering the HPV16 E6 protein and to C-terminal E7 peptides 108 and 109 (amino acids 62-80 and 72-97, respectively) in 49 women with biopsy-confirmed CIN. We have identified several E6 peptides containing human T-cell epitopes and again have demonstrated one or more reactive Tcell epitopes at the C-terminus of the HPV16 E7 protein ( 18 ) . Lymphoproliferative responses of PBMCs to E6 and E7 peptides correlated significantly with status of HPV infection, in that women who responded to E6 and/or E7 peptides were significantly more likely to be HPV negative than were nonresponders, suggesting that cell-mediated immune responses to HPV16 E6- and/or E7-transforming proteins are associated with resolution of HPV infection. Positive cell-mediated immune responses were more often seen with infection with HPV16 and closely related types than with other HPV types, suggesting some degree of HPV type specificity of cell-mediated immune responses, as has been previously reported by us and by other investigators ( 18 , 19 ) .
By prospectively testing a group of women who entered the study with biopsy-confirmed CIN grade II (moderate dysplasia), we have identified specific peptides in the E6 and E7 proteins to which cell-mediated immune responses correlated significantly with absence of HPV infection (presumed resolution) and with regression of CIN at subsequent clinic visits. Two peptides, one amino-terminal E6 sequence (amino acids 1-30) and one carboxyl- terminal E7 sequence (amino acids 72-97), to which cellmediated immune responses correlated significantly with favorable outcome were identified. Absent responses to these sequences correlated with persistent genital HPV infection as well as with persistent disease. Since previous studies have shown that HPV and its associated lesions tend to occur in synchrony ( 3 ) , it was not surprising to observe that positive cell-mediated immune responses to specific peptides had similar predictive value for HPV infection and for CIN status at the patient's next visit to the clinic, approximately 3 months later.
We have shown that some peptides (e.g., E6 amino acids 1-30 and E7 amino acids 72-97) have more protective activity than other peptides, perhaps because of the HLA types of the subjects tested or because of other factors. Early and still crude studies of reactive epitope mapping are presented and show that human T-cell epitopes are likely to be present within the Cterminal and N-terminal halves of E6 peptides 379 (amino acids 101-130) and 381 (amino acids 121-150) as well as within C-terminal peptide 370 (amino acids 16-30). An unexpected finding was that sequences present between amino acid 1 and amino acid 15 of E6 possibly interfere with processing or presentation of the reactive epitopes present between peptide 16 and peptide 30 when the complete longer peptide was tested.
When cervicovaginal lavage for HPV detection and blood sample for cell-mediated immunity studies were obtained from subjects at the same clinic visit (cross-sectional analysis), responders to one or more peptides were 3.1 times as likely as nonresponders to be negative for HPV infection ( P = .039); however, significant absence of CIN (regression) was not yet observed at this time. When positive cell-mediated immunity to any single peptide was detected by use of PBMCs obtained at one clinic visit, responders were 4.4 times as likely to have absent HPV infection by the next visit to the clinic (approximately 3 months later) as were nonresponders. When cellmediated immune responses to specific peptides (E7 peptide 109 and E6 peptide 369) were compared with nonresponses to these peptides, a significantly decreased prevalence of HPV infection and disease was observed within a 3-month period ( P = .021 and .018, respectively).
Twenty subjects had cell-mediated immune reactivity assessed at more than one clinic visit. Of these, nine (45%) were cell-mediated immunity positive to one or more peptides at the first visit. By the second testing, approximately 3 months later, 16 (80%) were responders and all of the nine subjects tested three or four times responded to one or more peptides by the fourth consecutive cell-mediated immunity assay. This result and the results described above suggest that development of cell-mediated immune responses to HPV-encoded epitopes and resultant resolution of infection take time after onset of HPVassociated disease. Whether responses to all reactive epitopes would correlate significantly with regression of HPV-associated disease, if enough subjects were studied, and whether cellmediated immune responses to specific peptides are associated with specific HLA types have yet to be determined.
Several of the patients studied here were participants in a β-carotene nutritional intervention clinical trial ( 3 , 33 ) . The likelihood of having cell-mediated immune reactivity to HPV16 peptides was significantly higher in the group given β-carotene (30-mg oral supplementation daily) than in the group given placebo (OR = 5.9; 95% CI = 2.5-13.7; P <.001). However, caution must be taken in interpreting these data, since this study of cell-mediated immunity was not designed to evaluate the effects of β-carotene, and only a small number of subjects (30 [30.6%] of 98) from the original clinical trial were studied for cellmediated immune responses. The difference in lymphoproliferative responses between the two treatment groups could well be due to extraneous factors, such as the fact that subjects in the β-carotene group underwent larger numbers of repeat cellmediated immunity tests and were thus more likely to ultimately test as cell-mediated immunity positive. Despite the increased prevalence of positive lymphoproliferative responses in the bcarotene group, it was observed that only six (24%) of 25 samples in the β-carotene group were HPV negative compared with three (43%) of seven cell-mediated immunity-positive samples in the placebo group. Although β-carotene may increase lymphoproliferative responses assessed in vitro , this finding does not necessarily imply that these responses are protective for HPV infection and its associated lesions in vivo . The β-carotene trial from which these subjects were taken did not find this antioxidant to be effective in causing regression of CIN ( 33 ) . Although inclusion of subjects from the β-carotene treatment group in this immunologic study might be a confounding variable, all results used in multivariate, longtitudinal analyses were controlled for randomization group (placebo or not).
Potential cytotoxic T-lymphocyte (CTL) epitopes identified in the HPV16 E6 and E7 proteins in a variety of murine model systems correlate well with the peptide sequences identified as “protective” in the human cell-mediated immunity experiments reported here. Both N-terminal E6 peptides and C-terminal E7 peptides have been predicted by several investigators to be potential CTL epitopes ( 34 , 35 ) , and experiments have suggested that E6 and E7 peptides are processed and expressed at the cell surface and are likely to be tumor rejection antigens ( 17 , 36 ) . By examining the abilities of specific peptide sequences to bind class I histocompatibility antigens, several groups ( 37 – 39 ) have identified E7 peptides capable of binding major histocompatibility complex class I molecules as potential CTL epitopes. Feltkamp et al. ( 40 ) used one high-affinity class I-binding peptide (E7 amino acids 49-57) to immunize mice. This immunization was sufficient to prevent growth of HPV16-transformed cell lines and to induce CTL in vitro . Using cloned CTL lines against this epitope, these investigators were able to eradicate established tumors in nude mice. They were thus able to demonstrate clearly that the E7 protein is a tumor rejection antigen and have established a murine model system for immunotherapy by adoptive transfer.
Other recent studies have evaluated human cell-mediated immune responses to HPV16 L1, E6, and/or E7 peptides by use of different experimental systems. Several groups ( 19 , 20 , 41 ) have used a “split-well” system in which PBMCs are first cultured with entire E6 or E7 fusion proteins and are then stimulated in short-term proliferation assays with peptides. Using pooled E7 peptides in split-well assays, de Gruijl et al. ( 19 ) found, as we have ( 18 ) , that detecting cell-mediated immune responses to E7 peptides in PBMCs required long culture periods (>2 weeks). The lengthy time required is most likely due to the small numbers of circulating PBMCs reactive with HPV antigens expressed in microscopic CIN lesions, and little or no reactivity can be detected in shorter (6-day) standard lymphoproliferation assays. Cell-mediated immune reactivity to peptides correlated with HPV types (greater in women infected with HPV16 than in HPV-negative subjects), but it did not correlate with seroreactivity to similar peptides, as we have reported ( 18 ) . However, de Gruijl et al. ( 19 ) found—quite differently from our results—that cell-mediated immunity to E7 peptides was more common in women with persistent HPV16 infection than in women with cleared or fluctuating infection. Explanations for the different results of the two studies are unclear at present, but they could be due to differences in study design. For example, patients in the study by de Gruijl et al. never had cervical biopsies, whereas our patients did, which could conceivably have affected cellmediated immune reactivity. Most of the patients with persistent infection in the study by de Gruijl et al. had CIN III, whereas most women in our study had CIN II. Although it is possible that results with pooled peptides as used in the study by de Gruijl et al. might have masked specific “protective” responses such as we have observed, the reasons for the discrepancy between those results and our results presented here are unclear.
Luxton et al. ( 20 ) found cell-mediated immune responses to N- and C-terminal E7 peptides among healthy control subjects more often than among subjects with cervical dysplasia, particularly to a peptide between amino acids 80 and 94 (similar to our peptide 109). More responses to E7 peptides were found among patients with high-grade CIN (CIN III) than among patients with low-grade lesions. Diminished cell-mediated immune responses to both E7 and L1 peptides were found among patients with cervical cancer, suggesting that patients with cancer may be nonresponsive to potentially protective epitopes. Using a bioassay for measuring antigen-stimulated cytokine production, Tsukui et al. ( 21 ) recently reported that IL-2 production by PBMCs stimulated for 6 days with pooled E6 and E7 peptides was greatest among normal subjects, lower among women with dysplasia, and still further decreased among women with highgrade CIN or cancer. Responses to C-terminal E6 and E7 peptides appeared to be common in all clinical groups. Of interest, these investigators did not observe IL-2 production stimulated by N-terminal E6 peptides, such as our peptide 369 (amino acids 1-30) which was highly correlated with favorable outcome in our study. Another recent study ( 42 ) has shown that non-HPVrelated recall antigen or allo-antigen-induced TH1 cytokines IL-2 and interferon were reduced in patients with CIN III with aggressive or extensive disease and that TH2 cytokines (IL-4 and IL-10) were increased compared with those in normal subjects and subjects with less aggressive disease, suggesting that these subjects have an underlying dysfunctional immune system.
To our knowledge, we have presented here for the first time data showing that human T-lymphocyte proliferative responses to HPV16 E6 and E7 peptides define a subset of patients with cervical neoplasia and that lymphoproliferative responses to specific E6 and E7 peptides are significantly associated with absence (presumed clearing) of HPV infection and with regression of CIN. Further studies in larger populations with long-term clinical, virologic, and immunologic follow-up must be done to establish definitively and to delineate the role of cell-mediated immune responses to specific HPV antigens in the natural history of HPV infection and HPV-associated neoplasia. Our results suggest that development of prophylactic vaccines and immunotherapies with the use of HPV16 E6 or E7 peptides or recombinant proteins might be feasible for prevention and treatment of human genital HPV infection and HPV-associated intraepithelial neoplasia and possibly cancer.