Classifying a cancer disease using CAIX expression

Abstract

The present invention relates to a carbonic anhydrase IX targeting compound for the use in the treatment of cancer, wherein the use comprises quantifying CAIX expression as well as the determination of a CAIX score based on the CAIX expression. The present invention relates further to a method for diagnosing, predicting and/or classifying a cancer disease comprising quantifying CAIX expression, and the determination of a CAIX score.

Claims

1. A method for classifying and treating a cancer disease comprising (a) obtaining a cancer sample from a cancer patient diagnosed with a non-metastatic type of cancer, (b) obtaining or establishing a score system for a specific staining agent and specific tumor cells using a reference tumor sample, by quantifying the expression of CAIX protein in said reference tumor sample in vitro by tumor cell staining, wherein a linear increase in staining of said reference tumor sample is determined and stained cells are divided into three groups of increasing staining wherein the respective increase in staining is the same for each group, wherein a fourth group has no staining and wherein said specific tumor cells are the same type of tumor cells as in said cancer sample, (c) quantifying the expression of CAIX protein in said cancer sample in vitro by tumor cell staining and determining a CAIX score by the formula: CAIX Score=(0percentage of viable tumor cells with no staining)+(1percentage of viable tumor cells with faint staining)+(2percentage of viable tumor cells with moderate staining)+(3percentage of viable tumor cells with strong staining), wherein faint staining, moderate staining and strong staining are the three groups with increasing staining determined based on said score system in (b), and wherein said tumor cells in said cancer sample are stained using the same specific staining agent as used in step (b), (d) comparing the CAIX Score with a predetermined threshold value of 2.0, (e) classifying the cancer disease as susceptible to treatment with a CAIX targeting compound if the CAIX score is 2.0, and (f) administering a CAIX targeting compound to said cancer patient classified as susceptible to treatment with a CAIX targeting compound, wherein said CAIX targeting compound is an anti-CAIX antibody or a functional fragment thereof which comprises the same six CDRs as the full length anti-CAIX antibody girentuximab.

2. The method according to claim 1, wherein said CAIX targeting compound is the anti-CAIX antibody comprising the same six CDRs as the full length anti-CAIX antibody girentuximab.

3. The method for classifying and treating a cancer disease according to claim 1, wherein the cancer is renal clear cell carcinoma.

4. The method according to claim 1, wherein said anti-CAIX antibody is chimeric G250 antibody or humanized G250 antibody.

5. The method according to claim 4, wherein said chimeric G250 antibody is girentuximab.

6. The method according to claim 1, wherein said CAIX targeting compound is the functional fragment comprises the same six CDRs as the full-length anti-CAIX antibody girentuximab.

Description

FIGURE LEGEND

(1) FIG. 1: HR (hazard ratio) and p-values for increasing CAIX-score values for the ITT (intended to treat) and PP populations (per protocol population)

(2) FIG. 2: DFS by treatment arm (cG250 vs. Placebo) for CAIX cut-off of 2.6

(3) FIG. 3: DFS curve for PP population with CAIX scores 2.6

(4) FIG. 4: High CAIX Score (2.0) and Age <65 years (ITT population)

EXAMPLES

(5) The following results are based on a randomised double blind Phase III study to evaluate adjuvant cG250 treatment versus placebo in patients with renal clear cell carcinoma (ccRCC) and high risk of recurrence.

(6) Objectives:

(7) The objectives in this study were to evaluate disease free survival (DFS) on girentuximab therapy as compared to placebo and to evaluate overall survival (OS) on girentuximab therapy as compared to placebo.

(8) Patients:

(9) 864 patients (ITT) were enrolled in a prospective, two armed, randomized, double blinded, placebo controlled study. The evaluation of the disease free survival was assessed by an independent Radiological Review Committee. 855 patients received at least one study treatment (cG250 or placebo) and were analysed for safety by an Independent Data Monitoring Committee. Inclusion criteria for the patients were: age 18 years, prior (partial or total) nephrectomy of primary renal cell carcinoma with documented clear cell histology, no evidence of macroscopic and microscopic residual disease. Patients should have one of the following High Risk Group (RG) (referring to TNM classification, 6th edition UICC, 2002): RG I: T3aN0/XM0 or T3bN0/XM0 or T3cN0/XM0 or T4N0/XM0 RG II: any T stage and N+ disease and M0 RG III: T1bN0/XM0 or T2N0/XM0, each with grading G 3 (Fuhrman or any other nuclear grading system with at least 3 grades),
not more than 12 weeks between date of nephrectomy and randomisation, ECOG of 0 to 1.
Study Medication:

(10) Girentuximab (cG250; an IgG1 kappa light chain chimeric version of an original murine monoclonal antibody mG250) was applied in a single loading dose of 50 mg (week 1) followed by weekly infusions of 20 mg of girentuximab (weeks 2-24). The drug was diluted in 100 ml normal saline (0.9% sterile sodium chloride in water) and was applied by intravenous infusion over 15 minutes. The placebo group received phosphate-buffered saline with polysorbate 20 (Tween 20) without active ingredient. Duration of treatment was 24 weeks.

(11) Immunohistochemistry:

(12) Paraffin embedded nephrectomized tissue specimens were cut into section of 3-5 m and collected on respective adhesive slides. Deparaffinisation and dehydration was done with EZ Prep mixture from Ventana Medical Systems, Inc. The tissue material was blocked and stained with a solution of the diluted antibody (CAIX M75, 1:150). The antibody was detected using a biotinylated IgG anti-mouse IgG antibody (Ventana Medical Systems, Inc.) followed by conjugated streptavidin horseradish peroxidase (Ventana Medical Systems, Inc.). The signal detection was done with diaminobenzidine and hydrogen peroxide (Ventana Medical Systems, Inc.). Positive and negative tissue controls were included as a reference.

(13) Statistical Methods:

(14) The efficacy evaluation was primarily analysed for the Intent-to-treat (ITT) population consisting of all 864 patients randomised. In addition the analysis was repeated for the Per Protocol (PP) Population consisting of 766 patients. Patients who received at least eight consecutive administrations of study medication (week 1 to 8) and had no major protocol deviation as defined in the deviation criteria prior to unblinding were evaluated as Per Protocol (PP) Population.

(15) The sample size calculation was performed using the program Pass 2002. Statistical summaries were produced using SAS software version 8.1 or higher. For the efficacy analysis, incomplete/partial dates were imputed; missing safety dates were not imputed. Adverse events and medical histories were coded using the Medical Dictionary for Regulatory Activities; medications were coded using the World Health Organisation Drug Dictionary. The primary efficacy analysis was based on the Intent-to-treat (ITT) Population and was repeated using the Per Protocol Population.

(16) Continuous variables were described using: number of observations, arithmetic mean, standard deviation, minimum, median and maximum. Categorical variables were presented using the number of observations and percentages. Hierarchical testing was applied for DFS and OS to keep the global significance level to 5% for both. Both DFS and OS were compared between the girentuximab arm and the placebo arm using the log-rank test and the Kaplan-Meier method. The significance levels for OS were adjusted using the O'Brien-Fleming approach for group sequential methods with an overall significance level of 5%. 95% confidence intervals for proportions were calculated using the exact method (Pearson-Clopper).

(17) The potential effect of prognostic factors on both DFS and OS was investigated using the Cox proportional hazards model. The hazard ratio (girentuximab treatment versus placebo) was estimated together with its associated 95% confidence interval and p-value using a proportional hazards model.

(18) Summary of Results:

(19) At the time of the cut-off for data analysis 360 patients (41.7%) had experienced a DFS event and 504 (58.3%) were still without disease according to local investigator assessment. 0.6% of the DFS patients were locally assessed as having metastatic disease at baseline. Overall 181 patients had died. After the IRRC central reading assessment, a DFS event could be assigned for 389 patients (45%) whereas a censor date could be assigned for the remaining 475 (55%) patients. The number of DES events (293 excluding patients with metastatic disease at baseline) that occurred in the ITT population was comparable between the treatment arms (cG250: 142, 32.8%; Placebo: 151, 35.0%), as was the rate of metastases at baseline (DFS=0), seen in 11.5% of girentuximab patients and 10.7% of placebo patients.

(20) The primary analysis for DFS based on the IRRC evaluation for the ITT population showed no statistically significant difference in overall median DFS between the treatment arms (hazard ratio [HR]=0.999, p=0.737). The median DFS for girentuximab was 71.4 months and the median DFS for the placebo group was not reached.

(21) There was no statistically significant difference in DFS between treatment arms, regardless of High Risk Group (RG) classification (RG I: HR=0.934, p=0.596; RG II: HR=1.73, p=0.084; RG III: HR=0.984, p=0.627).

(22) In addition, exploratory analyses done using a classification based on the CAIX antigen expression alone according to Bui et al. (Bui et al., Clinical Cancer Research 2003, 9: 802-811) have not delivered statistical significant results neither for the prognosis nor for the treatment effect. However, surprisingly the combination of the CAIX antigen expression with the intensity of the staining, the CAIX score, which can be considered as a measure of the antigen density, showed significant results for the treatment effect as well as for the prognosis.

(23) Multivariate analysis for the CAIX score cut-offs (1.91 for ITT and 1.52 for ITT Placebo) derived from survival tree analyses show that the CAIX score is a prognostic, but not independent factor for prognosis in DFS ITT, OS ITT and OS Placebo. The results suggest the CAIX score cut-off of 1.52 (ITT Placebo) may be useful for prognosis.

(24) For patients with a low CAIX score the Kaplan-Meier-Curves for the different treatment arms overlap independently of the cut-off used. This clearly shows that patients with a low CAIX score do not benefit from treatment.

(25) A high CAIX score of 1.91 showed a positive trend for the treatment effect (HR=0.879, p-value 0.248). Subgroup analyses for all CAIX scores from 0.0 to 3.0 revealed that as the CAIX score increases, the more significant the treatment effect becomes (FIG. 1).

(26) A CAIX score of 2.6 results in a statistically significant treatment effect with median DFS increasing from 51.2 months in the placebo arm to 73.6 months in cG250 patients (17% of the whole patient population, HR=0.568; p=0.022; FIG. 2). The 151 patients in the ITT population with a high CAIX score of 2.6 were evenly distributed between the cG250 and the placebo arm. Interestingly, patients in the placebo arm showed a 5 year DFS time (48.4 months versus 51.7 months) similar to patients with a CAIX score <2.6 regardless of treatment arm. Adjusting the population for patients with metastasis at baseline or for patients with positive lymph nodes did not change the outcome. Taking the PP population with a CAIX score 2.6 (N=139) improved both the hazard ratio and the significance level (HR=0.508; p=0.007; FIG. 2).

(27) Overall, the outcomes seen in the ITT population were reflected by the outcomes in the PP population (FIG. 3).

(28) A further subgroup analysis revealed a statistically significant treatment effect with median DFS increasing from 55.2 months in the placebo arm to 70.9 months in cG250 patients with a CAIX score of 2.0 and <65 years (HR=0.60; 95% CI 0.40-0.89; p-value=0.01; FIG. 4).

(29) In addition, the analyses have shown that clear cell RCC patients with a high T-stage (T3/T4), a low Grading (G1/G2), no lymph node involvement (N0/NX) and no metastatic disease have a significantly worse prognosis with a low CAIX score as compared to patients with a high CAIX score.