Cannabis-and-Cancer Research Is Advancing Through Personalized Screening

Cannabis research in oncology is moving beyond broad claims about “medical marijuana” and toward a more targeted question: could specific cannabinoid combinations be matched to individual tumors or cancer-related symptoms?

Israeli-based Cannabics Pharmaceuticals has pursued that approach through automated laboratory screening. The company’s platform tests cannabinoids and cannabinoid combinations against cancer cell lines and patient-derived samples, then uses imaging, high-throughput analysis and computational tools to measure how the cells respond. The long-term aim is to develop an ex-vivo sensitivity test that could help identify which compounds are most likely to affect a particular patient’s tumor.

That work is different from demonstrating that cannabis treats cancer in people. Laboratory findings can reveal whether a compound affects cultured cancer cells, but they do not establish that the same treatment will shrink tumors, extend survival or be safe and effective in patients. Cancer cells also differ substantially between tumor types—and between patients with the same type of cancer—making standardized conclusions difficult.

Cannabics’ human research has primarily involved symptom management rather than proving cannabis to be an anti-tumor treatment. In a 2019 pilot study published in Integrative Cancer Therapies, researchers evaluated controlled-release THC and CBD capsules in advanced cancer patients with cancer-related cachexia and anorexia syndrome. Of 24 people who consented, 17 began treatment and only six completed the six-month study. Three of the 17 patients who started treatment gained at least 10% of their baseline weight, while several others maintained their weight. Some participants reported improved appetite, but the study was small, uncontrolled and affected by withdrawals related to disease progression, chemotherapy side effects and cannabis-related symptoms.

The study therefore provides an early signal that standardized cannabinoid capsules might warrant further investigation for appetite loss and weight decline. It does not show that cannabis treated the participants’ cancers, and the authors called for larger studies.

The company’s separate screening program is designed to address the variability that makes cannabinoid research especially challenging. Automated microscopy and other high-content methods can record many cellular responses across different doses and cannabinoid ratios. Machine-learning tools may help researchers organize those results and identify patterns that would be difficult to detect manually. However, a computational model or cell-based test still requires independent validation before it can guide routine clinical care.

The broader evidence remains similarly cautious. The National Cancer Institute’s evidence summary on cannabis and cannabinoids distinguishes between potential use for symptoms—such as chemotherapy-related nausea, pain or appetite loss—and the much less established question of treating tumors themselves. Cannabis has not been approved by the U.S. Food and Drug Administration as a cancer treatment. Certain cannabinoid medicines, including dronabinol and nabilone, are approved for chemotherapy-induced nausea and vomiting, but that is not the same as an anti-cancer indication.

Regulatory requirements have also made cannabis studies more complicated in the United States, while countries such as Israel have provided research environments in which controlled cannabinoid studies could proceed. Cannabics’ original clinical trial is listed in ClinicalTrials.gov, offering a record of the study’s design and objectives.

Personalized screening may eventually help researchers determine which cannabinoid formulations deserve clinical testing and which patients are most likely to benefit. For now, though, the strongest conclusion is narrower: cannabis and individual cannabinoids remain promising subjects for laboratory and symptom-management research, but evidence that they can treat cancer itself is still preliminary.