Developing a Phosphospecific IHC Assay as a Predictive Biomarker for Topoisomerase I Inhibitors

Artificial intelligence scale-invariant feature transform algorithm-based system to improve the calculation accuracy of Ki-67 index in invasive breast cancer: a multicenter retrospective study

BACKGROUND

Ki-67 is a key indicator of the proliferation activity of tumors. However, no standardized criterion has been established for Ki-67 index calculation. Scale-invariant feature transform (SIFT) algorithm can identify the robust invariant features to rotation, translation, scaling and linear intensity changes for matching and registration in computer vision. Thus, this study aimed to develop a SIFT-based computer-aided system for Ki-67 calculation in breast cancer.

METHODS

Hematoxylin and eosin (HE)-stained and Ki-67-stained slides were scanned and whole slide images (WSIs) were obtained. The regions of breast cancer (BC) tissues and non-BC tissues were labeled by experienced pathologists. All the labeled WSIs were randomly divided into the training set, verification set, and test set according to a fixed ratio of 7:2:1. The algorithm for identification of cancerous regions was developed by a ResNet network. The registration process between paired consecutive HE-stained WSIs and Ki-67-stained WSIs was based on a pyramid model using the feature matching method of SIFT. After registration, we counted the nuclear-stained Ki-67-positive cells in each identified invasive cancerous region using color deconvolution. To assess the accuracy, the AI-assisted result for each slice was compared with the manual diagnosis result of pathologists. If the difference of the two positive rate values is not greater than 10%, it was a consistent result; otherwise, it was an inconsistent result.

RESULTS

The accuracy of the AI-based algorithm in identifying breast cancer tissues in HE-stained slides was 93%, with an area under the curve (AUC) of 0.98. After registration, we succeeded in identifying Ki-67-positive cells among cancerous cells across the entire WSIs and calculated the Ki-67 index, with an accuracy rate of 91.5%, compared to the gold standard pathological reports. Using this system, it took about 1 hour to complete the evaluation of all the tested 771 pairs of HE- and Ki-67-stained slides. Each Ki-67 result took less than 2 seconds.

CONCLUSIONS

Using a pyramid model and the SIFT feature matching method, we developed an AI-based automatic cancer identification and Ki-67 index calculation system, which could improve the accuracy of Ki-67 index calculation and make the data repeatable among different hospitals and centers.

Multicohort Retrospective Validation of a Predictive Biomarker for Topoisomerase I Inhibitors

PURPOSE

The camptothecin (CPT) analogs topotecan and irinotecan specifically target topoisomerase I (topoI) and are used to treat colorectal, gastric, and pancreatic cancer. Response rate for this class of drug varies from 10% to 30%, and there is no predictive biomarker for patient stratification by response. On the basis of our understanding of CPT drug resistance mechanisms, we developed an immunohistochemistry-based predictive test, P-topoI-Dx, to stratify the patient population into those who did and did not experience a response.

PATIENTS AND METHODS

The retrospective validation studies included a training set (n = 79) and a validation cohort (n = 27) of gastric cancer (GC) patients, and 8 cohorts of colorectal cancer (CRC) patient tissue (n = 176). Progression-free survival for 6 months was considered a positive response to CPT-based therapy. Formalin-fixed, paraffin-embedded slides were immunohistochemically stained with anti-phospho-specific topoI-Serine10 (topoI-pS10), quantitated, and analyzed statistically.

RESULTS

We determined a threshold of 35% positive staining to offer optimal test characteristics in GC. The GC (n = 79) training set demonstrated 76.6% (95% confidence interval, 64-86) sensitivity; 68.8% (41-88) specificity; positive predictive value (PPV) 92.5% (81-98); and negative predictive value (NPV) 42.3% (24-62). The GC validation set (n = 27) demonstrated 82.4% (56-95) sensitivity and 70.0% (35-92) specificity. Estimated PPV and NPV were 82.4% (56-95) and 70.0% (35-92) respectively. In the CRC validation set (n = 176), the 40% threshold demonstrated 87.5% (78-94) sensitivity; 70.0% (59-79) specificity; PPV 70.7% (61-79); and NPV 87.0 % (77-93).

CONCLUSION

The analysis of retrospective data from patients (n = 282) provides clinical validity to our P-topoI-Dx immunohistochemical test to identify patients with disease that is most likely to respond to topoI inhibitors.

Targeting the human MUC1-C oncoprotein with an antibody-drug conjugate

Mucin 1 (MUC1) is a heterodimeric protein that is aberrantly overexpressed on the surface of diverse human carcinomas and is an attractive target for the development of mAb-based therapeutics. However, attempts at targeting the shed MUC1 N-terminal subunit have been unsuccessful. We report here the generation of mAb 3D1 against the nonshed oncogenic MUC1 C-terminal (MUC1-C) subunit. We show that mAb 3D1 binds with low nM affinity to the MUC1-C extracellular domain at the restricted α3 helix. mAb 3D1 reactivity is selective for MUC1-C-expressing human cancer cell lines and primary cancer cells. Internalization of mAb 3D1 into cancer cells further supported the conjugation of mAb 3D1 to monomethyl auristatin E (MMAE). The mAb 3D1-MMAE antibody-drug conjugate (ADC) (a) kills MUC1-C-positive cells in vitro, (b) is nontoxic in MUC1-transgenic (MUC1.Tg) mice, and (c) is active against human HCC827 lung tumor xenografts. Humanized mAb (humAb) 3D1 conjugated to MMAE also exhibited antitumor activity in (a) MUC1.Tg mice harboring syngeneic MC-38/MUC1 tumors, (b) nude mice bearing human ZR-75-1 breast tumors, and (c) NCG mice engrafted with a patient-derived triple-negative breast cancer. These findings and the absence of associated toxicities support clinical development of humAb 3D1-MMAE ADCs as a therapeutic for the many cancers with MUC1-C overexpression.