A Phase I Trial of Vaccination of CA9-Derived Peptides for HLA-A24-Positive Patients with Cytokine-Refractory Metastatic Renal Cell Carcinoma

Renal cell carcinoma (RCC) is resistant to conventional treatment modalities such as chemotherapy and radiotherapy, and immunotherapy with biological response modifiers such as IFNs or interleukin-2 (IL-2) shows limited effects. Therefore, the treatment of RCC patients with disseminated disease remains challenging (1). New immunotherapy approaches (e.g., vaccination with tumor lysate–loaded autologous dendritic cells) have shown some promise but, particularly, the individualized production of dendritic cells in combination with low response rates has hampered wide acceptance of dendritic cell treatment. Nevertheless, these studies have clearly shown that vaccination can lead to impressive tumor regression in certain cases.

Recently, a large number of tumor-associated antigens and their peptides recognized by MHC class I–restricted CTLs have been identified in various malignancies (2–6) and used in vaccination studies (7–10). In search for RCC-associated peptides able to induce CTLs, we have identified three CA9 antigen-derived peptides, CA9p219-227, p288-296, and p323-331, and showed their ability to induce HLA-A24-restricted, CA9-specific CTLs (11). MN/CA9 was initially isolated as a tumor-associated antigen from HeLa cells (12) and belongs to the family of carbonic anhydrases catalyzing the conversion of carbon dioxide and water to carbonic acid, which is involved in various aspects of normal physiologic processes. After the molecular cloning of CA9, it has become clear that CA9 is expressed in many malignancies [e.g., cervical cancer (13), esophageal (14), colorectal (15), and lung carcinomas (16)]. There is compelling evidence that CA9 may serve as a prognostic marker in RCC (17) and CA9 may be involved in cancer progression (18). In previous studies, we have shown homogeneous CA9 expression in the majority of RCC cases: ∼90% of all RCC express CA9 and 99% of clear-cell RCC express CA9 whereas normal kidney tissue is CA9 negative (19). Based on the high, strong, and homogeneous expression in RCC and the limited expression in normal tissues, CA9 has been proposed as a promising target molecule for RCC therapy. Furthermore, vaccination with a CA9 mimic (anti-idiotype antibodies bearing the internal image of G250, recognizing CA9) resulted in antitumor responses (20), suggesting that a CA9 vaccination may be useful for active specific immunotherapy in RCC. In the present study, a phase I trial of peptide vaccination was carried out in patients with progressive cytokine-refractory metastatic RCC to assess both the toxicity and capability to induce immune responses of these three peptides.

Patients and eligibility. Patients with metastatic RCC and pathologically confirmed clear-cell RCC were candidates for this study. All patients had progression of disease after standard cytokine therapy such as IL-2 and/or IFN-α. Eligibility requirements included HLA-A24 positivity; age of ≥20 and ≤80 years; an Eastern Cooperative Oncology Group performance status of 0 or 1; granulocyte count ≥ 3,000/mm³; hemoglobin ≥ 10 g/dL; plates ≥ 100,000/mm³; bilirubin and creatinine equal to or less than the institutional normal limits; life expectancy ≥ 12 weeks; measurable or evaluable disease; no immunotherapy, chemotherapy, or radiotherapy within 4 weeks (washout for 4 weeks); and negative serologic tests for hepatitis B, hepatitis C, and HIV. Patients with serious illness or an active secondary malignancy were excluded. Other exclusion criteria also included existence of immunosuppressive or autoimmune disease or receipt of immunosuppressive drugs (e.g., steroids). All patients were informed of the investigational nature of the study and signed an informed consent in accordance with the institutional guideline. This study was approved by the Nara Medical University ethics committee. Each patient underwent a complete pretreatment clinical evaluation including a clinical history, physical examinations with assessment of performance status, laboratory studies, and measurements of radiographic studies [e.g., ultrasound sonography, computed tomography, magnetic resonance imaging, and radionuclide bone scan].

Patient background. From July 2002 to December 2003, 23 patients with cytokine-refractory metastatic RCC were enrolled at Nara Medical University Hospital. All patients previously underwent radical nephrectomy or needle biopsy of the primary tumor and had clear-cell RCC. In addition, CA9 expression was confirmed by immunohistochemistry or reverse transcription-PCR analysis as described elsewhere (18). The characteristics of the patients are summarized in Table 1. The median age of the patients was 63 years (range, 44-74 years) and the male-to-female ratio was 17:6. Three patients had a performance status of 1 and the remaining patients had a performance status of 0. Eleven patients had metastatic disease at surgery and 12 patients developed metastasis after radical surgery. The median time from completion of cytokine therapy is 4.87 ± 0.92 weeks.

CA9 peptide vaccines. Three CA9 peptides, CA9 p219-227 (EYRALQLHL), CA9 p288-296 (AYEQLLSRL), and CA9 p323-331 (RYFQYEGSL), were prepared under Good Manufacturing Practice conditions by Multiple Peptide Systems (San Diego, CA). All peptides have the ability to induce HLA-A24-restricted and tumor-specific CTL activity (11). Each peptide was dissolved in a small amount of 7% sodium bicarbonate solution (Myron, Ostuka Pharmaceuticals, Tokushima, Japan) and diluted with saline at a concentration of 4 mg/mL.

Study design. A skin test was done before each vaccination by intradermal injection of 10 μg of each peptide. Immediate and delayed-type hypersensitivity reactions were determined at 15 minutes and 24 hours after injection. A positive skin reaction was defined as >30 mm diameter erythema and induration, when saline was used as a negative control for assessment of the hypersensitivity. If immediate hypersensitivity was negative, the peptide was injected. Each of the three peptides was emulsified in an equal volume of incomplete Freund's adjuvant (Montanide ISA-51, Seppie, Paris, France) and injected s.c. into independent sites of the upper arm. Two vaccination regimens were used (Fig. 1). Briefly, patients received vaccinations every 2 weeks, and 1 week after the 3rd and/or 6th vaccination, the toxicity of each peptide was evaluated (Table 2). After confirming the safety of these peptides, the vaccination was continued every 2 to 4 weeks until death, intolerance, marked disease progression, major violations, or patients' withdrawal of consent. Study objectives included determination the safety of CA9 peptide vaccination, assessment of induced immune responses, and evaluation of antitumor clinical responses.

Clinical monitoring. Baseline studies included history, physical examination, blood examination, complete biochemical profiling, and chest X-rays. All patients were followed up until death due to disease, intolerance, or self-withdrawal. Toxicity assessments were done at least every 2 weeks using National Cancer Institute Common Terminology Criteria for Adverse Events v3.0. Clinical and laboratory assessments were checked at each visit. Clinical responses were assessed by computed tomography scan, magnetic resonance imaging, or X-ray examination of the tumor lesions at 2 to 4 weeks after the 6th vaccination and every 3 months thereafter. Responses of measurable lesions were evaluated according to the WHO Response Evaluation Criteria in Solid Tumors.

Immunologic monitoring. Peripheral blood was obtained from the patients pre- and post (every 6th)-vaccinations. Peptide-specific CTL precursors in peripheral blood mononuclear cells (PBMC) were detected by IFN-γ release assay as described elsewhere (21–23). Briefly, PBMCs (1 × 10⁵ per well) were incubated with 10 μmol/L of each peptide in 200 μL of RPMI/AIMV^· medium including 10% FCS and 100 units/mL of IL-2 and restimulated with the corresponding peptide (20 μmol/L) every 3 days for up to 12 days. On day 12, 24 hours after the last stimulation, the cells were harvested and cocultured with C1R-A24 cells preloaded with either the corresponding peptide or HIV peptide as negative control, respectively. After incubation for 18 hours, the supernatants (100 μL) were collected and the amount of IFN-γ was measured by ELISA. The data were considered positive (specific CTL induction) when the level of IFN-γ production in response to each corresponding peptide was significantly higher (P < 0.05) than that in response to HIV control peptide. All experiments were carried out in four different wells and in duplicate.

Peptide-specific immunoglobulin G (IgG) was measured by ELISA. In brief, wells of 96-well microtiter plates were coated with individual CA9 peptide or HIV control peptide (20 μg/well), blocked with Block Ace (Yukijirushi, Tokyo Japan), and incubated with serial diluted serum samples (1:100, 1:200, 1:400, 1:800). Subsequently, wells were incubated with rabbit anti-human IgG (Dako, Glostrup, Denmark), followed by goat anti-rabbit IgG-conjugated horseradish peroxidase-dextram polymer (Envision, Dako), and developed with tetramethylbenzidine substrate solution (KPL, Guildford, United Kingdom). Reactions were stopped with 1 mol/L phosphoric acid and absorbance was read at 450 nm. The absorbance values of each sample were determined as absorbance units per milliliter. To determine the cutoff value, sera from 12 healthy donors were measured for their reactivity to both corresponding peptides and an HIV peptide (negative control), and Δabsorbance (absorbance value in response to a corresponding peptide subtracted from the response to an HIV peptide at serum dilution of 1:100) was used to determine the presence of peptide-reactive IgG. The mean ± 2 SD of Δabsorbance in each peptide (CA9p219, p288, and p323) was 0.0128 ± 0.010, 0.012 ± 0.008, and 0.020 ± 0.014, respectively. The set cutoff values (mean ± 2 SD of Δabsorbance) were 0.028, 0.020, and 0.034, respectively.

Cytotoxicity against CA9-positive tumor cells in the peptide-reactive CTLs was investigated using a standard 6-hour ⁵¹Cr release assay. Briefly, PBMCs were cultured in RPMI 1640 and AIMV medium (Life Technologies, Gaithersburg, MD) with 10% FCS, 100 units/mL IL-2, and 1 mmol/L MEM nonessential amino acids. The cells were harvested on day 14. SW620 (HLA-A24+, CA9+), SKRC-44 (HLA-A24−, CA9+), CA9 peptide–pulsed C1R-A24 cells, phytohemagglutinin-activated T cells (HLA-A24+, CA9−), and HIV peptide–pulsed C1R-A24 were used as target cells. Target cells were labeled with ⁵¹Cr for 1 hour at 37°C and used for the assay. After selection of CD8-positive cells, effector cells were preincubated with unlabeled K562 cells to inhibit lymphokine-activated killer activity. ⁵¹Cr release was measured in the supernatant after 6 hours of reaction at different E/T ratios (5:1, 10:1, 20:1, 40:1). The percent of specific lysis was calculated as % lysis = (sample well Cr release − spontaneous release) / (detergent release − spontaneous release).

Statistical analysis. A two-tailed Student's t test was used throughout the study and P < 0.05 was considered statistically significant.

Adverse reactions. Twenty-two of 23 patients developed grade 1 or 2 local skin reactions with redness and swelling at the injection sites. Six patients experienced grade 1 or 2 fever but no medication was required. Grade 1 fatigue was observed in six patients. Grade 1 or 2 itching of the body was noted in 12 patients and, of these, four patients complained of rash, which resolved in a week without any medications. One patient (patient 2) occasionally developed transient grade 1 headache and diarrhea within 24 hours after CA9 vaccination, which resolved within 3 to 12 hours. No hematologic toxicity, cardiovascular, hepatic, or renal toxicities were observed during vaccination.

Cellular immune responses. We evaluated the cytotoxic activity of CTLs induced by CA9 peptide vaccination using IFN-γ release against C1R-A24 cells pulsed with each peptide, as well as a ⁵¹Cr release against CA9 expressing cancer cells as read-out. No peptide-specific CTL was detected in pre-vaccination PBMCs by the IFN-γ release assay. Post-vaccination PBMCs obtained from the patients every 3 months showed gradual increases of CA9p219 and/or p288 peptide–specific IFN-γ release whereas CA9 p323 did not induce IFN-γ release in any patients (Table 3). With the exception of five patients (patients 3, 13, 14, 17, and 21), 16 patients showed the induction of specific CTLs reactive to CA9p219 and/or p288 after the 6th or 9th vaccination as measured by the IFN-γ release assay. Kinetic studies of four representative patients (patients 1, 2, 4, and 5) are shown in Fig. 2 (P < 0.05). In these patients, CTL responses to CA9p219 and/or p288 were induced and the level of IFN-γ increased with the increased number of vaccination.

Because the IFN-γ release assays do not provide information about the lytic capacity of the induced CTL, and because peptide-specific CTL not recognizing the native antigen may be induced, we examined the specific cytotoxic activity of cultured PBMCs from 10 patients (patients 1, 2, 3, 4, 5, 8, 10, 11, 12, and 15). Specific HLA-restricted cytotoxicity against both SW620 cells and the corresponding peptide-pulsed C1R-A24 cells was observed in five patients (patients 1, 2, 4, 10, and 12; Fig. 3). PBMCs from three patients (patients 1, 2, and 8) showed significantly higher levels of HLA-A24-restricted cytotoxicity against SW620 cells in comparison with peptide-pulsed target cells. In the remaining four patients, peptide-specific or HLA-A24-restricted cytotoxicity was not observed (data not shown). Peptide-specific delayed-type hypersensitivity reactions were observed in eight patients. A summary is presented in Table 3.

Humoral immune responses. Peptide-specific IgG reactivity in pre- and post-vaccination plasma was measured by ELISA. Pre-vaccination, IgG reactive to CA9p219 or p288 was detected in 6 of 23 patients (patients 1, 2, 5, 6, 8, and 22). Longitudinal follow-up after initiation of vaccination revealed a gradual increase of anti–CA9 peptide IgG production in many cases. Results of levels of anti–peptide IgG in post-vaccination plasma (the 6th or 9th vaccination) are summarized in Table 3. IgG reactive to CA9p323 became detectable in all the patients. IgG specific to CA9p219 or p288 was detected in 15 patients (patients 1, 2, 3, 4, 5, 6, 8, 10, 11, 12, 15, 16, 18, 19, and 22) and 8 patients (patients 1, 4, 5, 10, 12, 15, 18, and 22), respectively. Examples of the development of anti-peptide IgG levels are shown in Fig. 4. In the majority of the tested patients, a rapid increase of anti-CA9p323 IgG titer, a relatively slow increase of anti-CA9p219 IgG titer, and a very slow increase of anti-CA9p288 IgG titer were found. It is of note that CA9p323 induced humoral responses, but not cellular responses, in all the patients tested, whereas both CA9p219 and p288 induced both humoral and cellular responses in selected patients.

Clinical responses. Of twenty-three patients, three patients (patients 1, 2, and 15) showed a partial response during CA9 vaccination therapy (Table 3). In the first patient (patient 1), slow progressive disease was observed in multiple lung metastases up to 6 months after the start of CA9 vaccinations, but ∼6 months later, the patient achieved a partial response with disappearance and shrinking of the lung metastatic lesions (Fig. 4). The second patient (patient 2) showed profound shrinking of pulmonary lesions and maintained in partial response for 11 months (Fig. 5). Unfortunately, this patient developed a brain metastasis and suddenly died due to unpredictable bleeding of the tumor. In the third patient (patient 15) also having multiple small lesions in the both lungs, several metastatic lesions disappeared and at present is stable (data not shown). Stable disease for >6 months was observed in six patients with a median duration of 12.2 months (7-18 months). The remaining 14 patients did not show any clinical responses.

At the time of the analysis in the present study, 13 patients have died of RCC and 10 patients were alive with a median follow-up of 14 months (5-35 months) and the median overall survival was 21.0 months.

Recent progress in the treatment of malignant tumors has resulted in a substantial improvement in the disease prognosis. However, the circumstances for patients diagnosed with metastatic RCC have not changed dramatically: these patients still have poor prognosis; their clinical course is unpredictable; and standard treatments such as chemotherapy and radiotherapy are ineffective and immunotherapy with biological response modifier such as IL-2 and IFNs have limited effect.

Several clinical trials of peptide-based trials have been done in recent years. In this study, we describe a clinical trial investigating CA9-derived peptide vaccination in HLA-A24-matched progressive metastatic RCC patients. Previous studies eluded that these peptides might be promising targets for RCC immunotherapy in HLA-A24-matched patients. This HLA-A24 genotype is predominant in Japanese (∼60%) and less abundant in Caucasians (17%), Blacks (9%), and Hispanics (27%; ref. 24). The primary end point was to evaluate the safety and toxicity of CA9 peptide vaccination. Only mild toxicity was observed and, therefore, this treatment is certainly acceptable for cytokine-refractory RCC patients. Importantly, gastrointestinal hepatic toxicity was not observed despite CA9 expression in large bile duct epithelium and gastric mucosal cells (i.e., there was no clinical evidence of autoimmune reactions; ref. 25).

The secondary aim of our study was to evaluate vaccine-induced specific immune reactions. Immunologic monitoring included IFN-γ release assays in response to individual peptides, cytotoxicity assays against CA9-positive cancer cells, measurement of serum IgG reactive with individual peptides, and measurements of specific delayed-type hypersensitivity skin reactions. We were able to show that CA9 peptide vaccination induced CA9p219 and/or CA9p288 peptide–specific CTLs in most patients whereas CA9p323-specific CTLs were not detected in any patient, suggesting that CA9p323 failed to induce CTLs in RCC patients. Of note, CA9/HLA-A24–positive cells were very efficiently recognized and killed, strongly suggesting that endogenously processed CA9 peptides (CA9p219 and CA9p288) were presented and served as TCR binding motifs. On the other hand, CA9p323 vaccination induced rapid and high production of the anti-CA9p323 antibody (IgG) in all evaluable patients, suggesting that this peptide is a B-cell target. Cytotoxicity assays revealed powerful CA9-specific, HLA-restricted lytic activity against RCC cells, showing that peptide vaccination resulted in the induction and expansion of antigen-specific CTLs. Although cellular immune responses have been suggested to play a major role in tumor regression, the role of (tumor-specific) humoral responses is unresolved but may be beneficial. Several investigators have reported that induction of peptide-specific IgG correlated with clinical response or survival of patients with prostate (26), gastric (27), lung (28), or gynecologic cancer (29). Mine et al. (30) reported that increased peptide-specific humoral responses, but not cellular responses, correlated with overall survival. Whether successful peptide vaccination requires induction of specific cellular as well as humoral immune responses is unclear.

Whereas the studied patient group is relatively small, it seems that the induction of both peptide-specific CTLs and IgG against CA9p219 as well as p288 is important. Of nine patients with a favorable disease course (i.e., three partial response and six stable disease), seven patients induced both CA9p219/p288–specific CTLs and IgG. In the remaining two cases, either CA9p288-specific CTLs or CA9p288-specific IgG was absent. A positive correlation between induction of IgG specific to vaccinated peptides and clinical response or survival of patients has been described for various tumor types (26–30). Our results also suggest a positive correlation between the induction of IgG reactive to CA9 peptides and prolonged survival. Of 11 patients not showing CA9p219 and/or p288-specific IgG, 10 showed progressive disease with short survival (median survival, 13.3 months). In a recent study, glioma patients received “personalized” peptide vaccines based on pre-vaccination PBMC peptide reactivity, with the assumption that immune boosting might be possible versus immune priming of naïve T cells (31). In our hands, pre-vaccination PBMC did not seem to contain any CA9 peptide–specific CTLs, as judged by IFN-γ release assays, whereas CTLs with strong lytic capacity were retrieved post-vaccination. However, peptide-specific tetramer staining did reveal the presence of specific CTL precursors in pre-vaccination PBMC from some, but not all, patients (data not shown). Thus, the presence of pre-vaccination CTL precursor may, in part, account for our success in inducing CA9 peptide–specific CTLs.

Mulders et al. (32) recently reported the results of a similar clinical trial using HLA-A2-restricted CA9 9-mer peptides for RCC patients. Six patients with disseminated RCC received vaccinations of dendritic cells pulsed with CA9 peptides and keyhole limpet hemocyanin. Humoral responses against keyhole limpet hemocyanin and delayed-type hypersensitivity conversion were induced in all patients; however, potent immune or clinical responses were not observed. This contrasts starkly with our study revealing induction of peptide-specific CTLs with lytic capacity in ∼80% of the patients, development of peptide-specific IgG responses, and clinical responses. The studies are difficult to compare because different peptides were used [i.e., HLA-A2 versus HLA-A24 vaccination consisted of CA9 peptide–loaded dendritic cells versus peptide with adjuvant (incomplete Freund's adjuvant)] and the number of vaccinations differed substantially. Nevertheless, the HLA-A24-restricted CA9 peptides may be more immunogenic than the HLA-A2-restricted CA9 peptides. It is noteworthy that induction of specific immune and clinical responses seemed to correlate with the number of vaccinations. More than 10 vaccinations were needed to obtain specific cellular responses (Table 3). This may partly be due to the use of peptide in combination with a nonnatural adjuvant, a combination which may not be very immunogenic, albeit that peptide vaccination of melanoma patients resulted in induction of peptide-specific CTLs. The median number of vaccinations in all patients was 24, but for patients without induction of either CA9p219/CA9p288-specific CTL or IgG (patients 3, 6, 9, 13, 14, 17, 20, and 21), the median was much lower (7-28; median, 14). It is unclear whether inductions of specific immune responses and clinical responses correlate. It has recently been suggested that peptide vaccination may lead to tumor destruction resulting in the expansion of CTLs directed against other peptides (33). Our findings suggest that continued vaccinations may lead to induction of specific immunity as well as prolonged survival.

In conclusion, treatment of cytokine-refractory progressive RCC patients with HLA-A24-restricted CA9 peptide vaccines was well tolerated and induction of antigen-specific immunity was observed in most of the patients. Considering the absence of effective treatment strategies for cytokine-refractory RCC patients, the observed clinical responses are encouraging, although patient bias may have played a role. The promising clinical responses warrant further exploration of CA9 peptides as an active specific immunotherapy for RCC.

We thank Dr. Akihisa Yao of Kobe University for technical support in CTL assays.