Abstract P3-05-10: The development of a standardized Ki-67 assay for the ALTERNATE trial: An experience in academic investigational device development

Autor: D. C. Allred, Cynthia X. Ma, Erika C. Crouch, Yu Tao, Vera J. Suman, Rodrigo Franco Gonçalves, S Sanati, Katherine DeSchryver, MJ Ellis, Michael Barnes
Rok vydání: 2013
Předmět:
Zdroj: Cancer Research. 73:P3-05
ISSN: 1538-7445
0008-5472
DOI: 10.1158/0008-5472.sabcs13-p3-05-10
Popis: Ki-67 proliferation marker-based response monitoring will be prospectively tested in the ALternate Treatment for ER positive NeoAdjuvant TrEatment (ALTERNATE) trial. Tumor Ki-67 will be analyzed after one month of neoadjuvant anastrozole, fulvestrant or the combination to detect early resistance (Ki-67 >10%) for triage to alternate treatment (chemotherapy/investigational agent). Ki-67 will be repeated on the surgical specimen after 24 weeks to detect PEPI-0 cases (path Stage 0/1/2A and Ki-67 ≤2.7%), which will be managed without adjuvant chemotherapy. A CLIA-approved Ki-67 assay was required by CTEP/FDA. FDA reviewed-components included the CONFIRM anti-Ki-67 (30-9) Rabbit Monoclonal Primary Antibody, the Benchmark XT platform and the VENTANA iScan Coreo scanner. The validation process was as follows: Step 1 was a TMA-based bridging study between a SP6-based research assay to the 30-9 CLIA assay using visual point counting (VPC); Step 2 trained a scoring SOP using Companion Algorithm Ki-67 (30-9) image analysis software with triage to VPC when certain observations were flagged; Step 3 was an independent test of the SOP using relevant patient outcome-associated samples from the P024 and POL neoadjuvant endocrine studies with two independent pathology analyses. STEP 1: Bridging Research to CLIA. Using a 28-case TMA, and VPC, the intra-class correlation (ICC) for agreement measure of Ki-67 10% cut-point (CP) between research Ki-67 and CLIA Ki-67 was 0.96 (CI 0.92-0.98). The Cohen's simple kappa for agreement measure was 0.78 (CI 0.55-1.0). STEP 2: Scanner-based Algorithm training. An SOP for the image analysis was developed by reviewing discordant cases in two independent reads at the 10% and 2.7% CP in a training set (N+ disease from P024 N = 61). Histology features that led to inaccurate scoring from the image analysis included: streaming /fascicular tumor cells, a large amount of intervening stromal cells, heavy immune cell infiltration, sparse scattered tumor cells, faint staining, and plump stromal cells. These cases were flagged for VPC. The final Ki-67 scoring SOP used image analysis with triage to VPC when necessary (required in 16% of cases). STEP 3: Scoring Algorithm Validation. The Ki-67 SOP was run on T1/2 N0 cases from the P024 trial N = 58. The% positive agreement for the combined scanner and VPC approach using the Ki-67 2.7% CP was 87.5% (95% CI 61.7- 98.5%); percent negative agreement 88.9% (95% CI: 65.3-98.6%). Minor discordance did not affect the ability to predict very similar survival outcomes (Log Rank P = 0.044 and P = 0.055). The data for the 10% CP was similar (N = 66), the percentage positive agreement was 100%; percent negative agreement 93.55% (95% CI: 78.58-99.21%). The survival predictions were again similar (P = 0.0001 and P = 0.0002). The FDA ruled the use of Ki-67 in the ALTERNATE trial is “no significant risk” because previous clinical trial data indicated the majority of patients with Stage 0/1/2A disease post neoadjuvant endocrine therapy already do not receive chemotherapy. Thus the use of Ki-67 reduces the risk of under-treatment of patients with low stage disease but aggressive biology. CTEP accepted these data on assay concordance and performance and the ALTERNATE trial will proceed. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P3-05-10.
Databáze: OpenAIRE
Externí odkaz: