Histology based classifier to personalize treatment for patients with colon cancer
Approximately one-third of colon cancer patients are diagnosed with early-stage disease, which amounts to over 3,000 patients per year in the Netherlands. The standard treatment for this is surgical resection, followed by optional additional chemotherapy to reduce the risk of recurrence. Currently, oncologists are unable to predict which patients require this additional chemotherapy, and for which patients surgery alone is sufficient for a cure. The UMC Utrecht and DoMore Diagnostics have initiated this partnership to do predict this using the Histotype Px Colorectal artificial intelligence (AI) based digital biomarker with the ultimate goal to spare nearly half of the patients unnecessary chemotherapy.
CRC incidence is expected to rise 30.3% globally (2020-2030), so there is a growing need to stop non-beneficial treatment. The proposal aims to deliver new knowledge to switch from a one-size fits all approach to personalized treatment. The aim is to provide a biomarker that oncologists can use to select only patients with an expected benefit for chemotherapy treatment in order to increase their chance of cure and on the other hand spare the 40% of patients with expected low risk unnecessary treatment to spare these patients the associated toxicity. Our innovation is aimed to enable tailored, patient-centric cancer care that minimizes potential side effects and maximizes treatment effectiveness, quality of life and healthcare resource use.
We will use of standard tumor tissue and clinical data from the Netherlands Cancer Registry. Using digitalized slides DoMore Diagnostics will apply their biomarker to classify patients into risk groups. This will be done on a matched cohort of patients treated with and without adjuvant chemotherapy. We will evaluate whether the low risk patients will have already an excellent prognosis without adjuvant chemotherapy and can be spared treatment with chemotherapy. Furthermore we will use these data to learn which patients do and do not benefit from adjuvant
chemotherapy in order to further personalize treatment. The eventual goal is to deliver a prognostic and predictive biomarker which can be used in daily clinical practice to guide treatment decision making in order to increase cure rates, survival, quality of life, and reduce unnecessary health care costs and side effects.
Current guidelines recommend adjuvant chemotherapy after surgery to all patients with a locally advanced colon tumor. However, half of the patients will already be cured by the surgery alone and can be safely omitted chemotherapy. When a patient with low risk of recurrence can be identified upfront, adjuvant chemotherapy can be safely spared, reducing unnecessary toxicity risks and costs for society. Therefore, the biomarker Histotype Px was developed. Histotype Px is a tissue based biomarker, using artificial intelligence (AI) to predict prognosis based on tumor tissue. It has been further developed by integrating standard pathological features (TNM stage and the number of resected lymph nodes) into a combined model called CAPAI (Combined Analysis of Pathology and Artificial Intelligence). With this project we aimed to validate this biomarker CAPAI in the Dutch population.
According to national guidelines, adjuvant chemotherapy is recommended for these patients, however, some patients chose not to receive it. This enables comparison of outcomes between patients treated with and without adjuvant chemotherapy. This approach allows validation of the prognostic value of CAPAI in both groups, as well as assessment of the potential benefit of chemotherapy. Therefore, surgical tissue samples were collected from all patients in the Netherlands who underwent surgery for locally advanced colorectal cancer between 2015 and 2019 that did or did not receive adjuvant treatment. After collecting all tissue samples, they were digitalized, and classified by CAPAI into low-, intermediate- or high risk. In addition, clinical data was gathered from the Netherlands Cancer Registry.
As a first step, we analyzed results from 453 patients who did not receive adjuvant chemotherapy. The biomarker CAPAI stratified patients in distinct prognostic risk groups, stratifying most patients into the low-risk group (n = 215, 47%), followed by the intermediate-risk (n = 156, 34.4%), and high-risk group (n = 82, 18.1%). Low-risk patients showed a 3-year cancer-specific survival rate of 93.7%, demonstrating that the biomarker can identify a substantial subgroup of patients with an favorable prognosis, without receiving adjuvant chemotherapy.
The next step as part of the Health Holland follow up project, is to identify low-risk patients who did receive adjuvant chemotherapy, to enable matched comparison between the treated and untreated group. If low-risk patients demonstrate similar outcomes regardless of adjuvant chemotherapy, it may indicate that chemotherapy can be safely omitted in this CAPAI defined subgroup.
This way, the biomarker CAPAI provides an important step toward more personalized treatment for patients with colorectal cancer.