SP Somashekhar, MBBS, MS, MCH, FRCS

The artificial intelligence computer program Watson for Oncology (WFO) achieved a high degree of concordance with tumor board recommendations in a double-blinded validation study in Bengaluru, India, according to results presented at the 2016 San Antonio Breast Cancer Symposium (SABCS).

In the study of cases involving 638 patients with breast cancer treated at Manipal Comprehensive Cancer Center, 90% of WFO’s recommendations for standard treatment (REC) or consideration (FC) were concordant with the recommendations of the tumor board. A group of 12-15 oncologists met weekly to review cases, entered data into the WFO system, and then analyzed the degree of concordance between WFO’s recommendations and those of the tumor board, as well as the time it took the oncologists to generate their recommendations.

The degree of concordance varied according to the type of breast cancer, lead study author SP Somashekhar, MBBS, MS, MCH, FRCS, said in his presentation at SABCS. WFO recommendations were concordant nearly 80% of the time in nonmetastatic disease, but only 45% of the time in metastatic cases. In cases of triple-negative breast cancer, WFO agreed with physicians 68% of the time, but in HER2/neu-negative cases, WFO’s recommendations matched the physicians’ recommendations only 35% of the time.

In cases of discordance between WFO’s recommendations and those of the tumor board, tumor board decisions were changed 63% of the time (n = 100) following review.

The study’s authors concluded that the broader divergence between WFO’s recommendations and those of the tumor board could be attributed to the greater number of treatment options available for patients with HER2/neu-negative breast cancer.

“Including HER2/neu cases opens up many more treatments and variables for consideration,” Somashekhar, chairman of the Manipal Comprehensive Cancer Center, explained. “This increases demands on human thinking capacity. More complicated cases led to more divergent opinions on the recommended treatment.”

Physicians took longer to weigh the available treatment options and come to a recommendation compared to WFO, although the doctors were able to work faster as they gained familiarity with cases. Somashekhar said it took doctors an average of 20 minutes initially. As they improved, the mean time dropped to about 12 minutes. By comparison, WFO achieved a median time of 40 seconds to capture and analyze data and give a treatment recommendation.

The study authors said that although WFO recommendations often led the tumor board to reconsider their decisions, the computer program remains a support tool for physicians and cannot replace the “human touch” needed to act upon the many factors of patient engagement that go beyond data analysis.

“We are dealing with human beings, and the context and preferences of each individual patient, the patient—physician relationship, and the human touch and empathy are very important,” Somashekhar said. “It’s always going to be the decision of the treating oncologist and patient to determine what is truly the best option for the patient.”

One reason why physicians do not have to worry about Watson replacing them is that they can perform better at 1-on-1 assessments. For example, whereas in metastatic disease, Watson tended to recommend conservatively based on best available evidence, physicians were more likely to select an aggressive chemo regimen to achieve a high level of response, Somashekhar said. This explains some of the discordance in recommendations, he added.

In the study, WFO analyzed >100 patient attributes for breast cancer and provided a ranking of treatment options according to REC, FC, and “not recommended” (N-REC). The recommendations were backed by data from recent trials, and oncologists were able to click on options listed by Watson to find out more about the recommendations and the reasons for them. The cases were at most 3 years old.

Somashekhar said the study was not designed to evaluate why differences in recommendations occurred, the inferiority or superiority of recommendations, or the impact of WFO on workflow. He said WFO, developed by IBM, is a promising tool that warrants consideration in a variety of other clinical settings and study designs.

Doctors at Memorial Sloan Kettering Cancer Center (MSK) helped to program WFO to enable it to make recommendations on cancer treatment. The system extracts and assesses large amounts of structured and unstructured data from medical records through natural language processing and machine learning. In addition to breast tumors, it is also capable of making recommendations for lung and colorectal cancers.

WFO’s concordance with MSK oncologists’ opinions has been tested in 2 previous studies, showing agreement 90% of the time in one and 50% of the time in another. Doctors in Thailand have been using the system for more than a year, and IBM announced this past summer that it was expanding the program to China, where it was expected to be of high value to doctors in rural centers who don’t have access to resources available to doctors in centralized clinics.

Watson, which has been in use in the Manipal hospital system for 6 months, has proven valuable in controlling cancer clinic costs, because it helps to eliminate bias and errors. “This is something that would ensure that we arrive at the right decision first,” Somashekhar said.

Somashekhar SP. Double blinded validation study to assess performance of IBM artificial intelligence platform Watson for oncology in comparison with Manipal multidisciplinary tumor board—first study of 638 breast cancer cases. Presented at: San Antonio Breast Cancer Symposium, Friday, Dec. 9, 2016; San Antonio, TX. Abstract S6-07

<<<

View more from the 2016 San Antonio Breast Cancer Symposium

Image Credit: © Sebastian Kaulitzki
– stock.adobe.com

The addition of inavolisib (Itovebi) to palbociclib (Ibrance) and fulvestrant (Faslodex) led to a statistically significant improvement in overall survival (OS) vs placebo plus palbociclib and fulvestrant in patients with PIK3CA-mutant, hormone receptor–positive, HER2-negative, endocrine-resistant advanced breast cancer, according to data from the final OS analysis of the phase 3 INAVO120 trial (NCT04191499) that were presented in a press briefing ahead of the 2025 ASCO Annual Meeting.1

At a median follow-up of 34.2 months, the median OS was 34.0 months (95% CI, 28.4-44.8) with inavolisib (n = 161) vs 27.0 months (95% CI, 22.8-38.7) with placebo (n = 164; stratified HR, 0.67; 95% CI, 0.48-0.94; P = .0190). The 6-, 12-, 18-, 24, and 30-month OS rates in the inavolisib arm were 96.8%, 87.0%, 74.3%, 65.8%, and 56.5%. The respective rates in the placebo arm were 90.1%, 76.7%, 67.2%, 56.3%, and 46.3%.

“This is the first time OS has been significantly improved by a PI3K pathway targeted drug,” Nicholas C. Turner, MD, PhD, lead study author and director of The Royal Marsden and Institute of Cancer and National Institute for Health and Care Research Biomedical Research Centre in London, United Kingdom, said in the press briefing.

INAVO120 Trial Background and Grounds for Approval

Mutations in the PIK3CA gene are present in approximately 40% of advanced hormone receptor–positive, HER2-negative breast cancers and confer poor prognosis and response to PI3K inhibitors. Key components of tumor growth involve the estrogen receptor, CDK4/6, and PI3K signaling pathways. As such, inhibition of all three pathways can augment treatment response and prolong the time to resistance. However, previous attempts to combine PI3K inhibitors and CDK4/6 inhibitors have been futile because of toxicity.

Inavolisib is an oral, highly potent, selective PI3K inhibitor that causes degradation of the mutated PI3Kα, p110α, and can be safely combined with palbociclib and fulvestrant at full dose.

This triplet regimen has been approved by the FDA since October 2024, based on earlier findings from INAVO120 showing a more than doubling in the median progression-free survival (PFS) and overall response rate (ORR) vs palbociclib and fulvestrant alone.2

INAVO120 was a double-blind trial that enrolled patients with PIK3CA-mutant, hormone receptor–positive, HER2-negative advanced breast cancer by central circulating tumor DNA (ctDNA) or local tissue/ctDNA confirmation.1 To be eligible for enrollment, patients also needed to have measurable disease and disease progression during or within 12 months of completing adjuvant endocrine therapy. Prior therapy for advanced disease was not permitted, nor was fasting glucose levels above 126 mg/dL and hemoglobin A1C levels above 6%.

Enrollment took place between January 2020 and September 2023, during which 325 patients were randomly assigned 1:1 to the inavolisib or placebo arm. In the investigational arm, patients received 9 mg of oral inavolisib daily plus 125 mg of oral palbociclib on days 1 through 21 and 500 mg of fulvestrant on days 1 and 15 of cycle 1 and every 4 weeks thereafter. In the control arm, patients received placebo in place of inavolisib plus the same dose and schedule of palbociclib and fulvestrant. Treatment was continued until progressive disease or toxicity.

Patients were stratified by visceral disease (yes vs no), endocrine resistance (primary vs secondary), and region (North America/Western Europe vs Asia vs other).

The primary end point was investigator-assessed PFS. Secondary end points were OS, investigator-assessed ORR, best overall response, clinical benefit rate, duration of response, and patient-reported outcomes.

Inavolisib Shines With Maintained PFS Benefit, Time to Subsequent Therapy

With an additional 12.8 months of follow-up investigators showed that the PFS benefit with inavolisib was upheld.1,3 The median PFS was 17.2 months (95% CI, 11.6-22.2) with inavolisib vs 7.3 months (95% CI, 5.9-9.2) with placebo (stratified HR, 0.42; 95% CI, 0.32-0.55).1 The 6-, 12-, 18-, and 24-month PFS rates with inavolisib were 83.4%, 58.0%, 49.7%, and 41.8%, respectively. The respective rates in the placebo arm were 57.9%, 31.3%, 20.5%, and 16.7%.

Additional efficacy data revealed that the time to first subsequent chemotherapy was substantially delayed with the addition of inavolisib. The median time to first subsequent chemotherapy was 35.6 months (95% CI, 25.4-not reached) with inavolisib vs 12.6 months (95% CI, 10.4-16.1) with placebo (stratified HR, 0.43; 95% CI, 0.30-0.60).

“I’m impressed with this study by the almost 2-year prolongation in the delay in going on chemotherapy from 12.6 months to 35.6 months. Delaying the need in the metastatic setting to go on chemotherapy by almost 2 years is certainly an outcome that matters to patients,” Julie Gralow, MD, chief medical officer and executive vice president at ASCO, remarked during the briefing.

Breaking Down the Safety Data

With respect to safety, all patients in both arms experienced an adverse effect (AE). The rates of grade 3/4 AEs in the inavolisib and placebo arms were 90.7% and 84.7%, respectively. Six patients (3.7%) in the inavolisib arm and 2 (1.2%) in the placebo arm had grade 5 events, none of which were deemed treatment related by investigators. Serious AEs occurred in 27.3% and 13.5% of patients in the inavolisib and placebo arms, respectively.

AEs leading to discontinuation in the inavolisib and placebo arms, respectively, occurred with inavolisib/placebo (6.8%; 0.6%), palbociclib (6.2%; 0%), and fulvestrant (3.7%; 0%).

AEs leading to dose reduction in the inavolisib and placebo arms, respectively, occurred with inavolisib/placebo (14.9%; 3.7%), and palbociclib (40.4%; 34.4%).

Turner noted that the AEs were similar to prior reports and that inavolisib had a low treatment-related discontinuation rate.

“Inavolisib plus palbociclib and fulvestrant significantly improved OS compared with placebo plus palbociclib and fulvestrant in patients with PIK3CA-mutated, hormone receptor–positive, HER2-negative, endocrine-resistant advanced breast cancer,” Turner said in conclusion.

Disclosures: Turner listed no disclosures.

References

1. Turner N, Im SA, Saura C, et al. INAVO120 phase III trial final overall survival (OS) analysis of first-line inavolisib (INAVO)/placebo (PBO) + palbociclib (PALBO) + fulvestrant (FULV) in patients (pts) with PIK3CA-mutated, hormone receptor-positive (HR+), HER2-negative (HER2–), endocrine-resistant advanced breast cancer (aBC). Presented at: 2025 ASCO Annual Meeting; May 30-June 3, 2025; Chicago, IL. Abstract 1003.
2. FDA approves inavolisib with palbociclib and fulvestrant for endocrine-resistant, PIK3CA-mutated, HR-positive, HER2-negative, advanced breast cancer. FDA. October 10, 2024. Accessed May 21, 2025. https://www.fda.gov/drugs/resources-information-approved-drugs/fda-approves-inavolisib-palbociclib-and-fulvestrant-endocrine-resistant-pik3ca-mutated-hr-positive
3. Turner NC, Im SA, Saura C, et al. Inavolisib-based therapy in PIK3CA-mutated advanced breast cancer. N Engl J Med. 2024;391(17):1584-1596. doi:10.1056/NEJMoa2404625

First-Line Maintenance Lurbinectedin

Plus Atezolizumab in ES-SCLC | Image

Credit: © catalin – stock.adobe.com

Frontline maintenance treatment with lurbinectedin (Zepzelca) plus atezolizumab (Tecentriq) improved progression-free survival (PFS) and overall survival (OS) vs atezolizumab alone in patients with extensive-stage small cell lung cancer (ES-SCLC), according to primary results from the phase 3 IMforte trial (NCT05091567) presented at the 2025 ASCO Annual Meeting.1

The median PFS with the doublet (n = 242) was 5.4 months (95% CI, 4.2-5.8) by independent review facility (IRF) assessment vs 2.1 months (95% CI, 1.6-2.7) with the monotherapy (n = 241), translating to a 46% reduction in the risk of disease progression or death (HR, 0.54; 95% CI, 0.43-0.67; 2-sided P < .0001). The 6-month IRF-PFS rates in the respective arms were 41.2% and 18.7%; at 12 months, these rates were 20.5% and 12.0%.

A clinically meaningful OS benefit was also observed with the addition of lurbinectedin to atezolizumab vs atezolizumab alone, at a median of 13.2 months (95% CI, 11.9-16.4) vs 10.6 months (95% CI, 9.5-12.2), respectively (HR, 0.73; 95% CI, 0.57-0.95; 2-sided P = .0174). The 12-month OS rate with the doublet was 56.3% vs 44.1% with the monotherapy.

“IMforte is the first phase 3 study to show PFS and OS improvement with first-line maintenance treatment for ES-SCLC, highlighting the potential of lurbinectedin plus atezolizumab to become a new standard of care [SOC] for first-line maintenance therapy in patients with this aggressive and difficult-to-treat disease,” Luis Paz-Ares, MD, PhD, of Hospital Universitario 12 de Octubre, H12O-CNIO Lung Cancer Unit, Universidad Complutense and Ciberonc, in Madrid, Spain, said in a press briefing ahead of the meeting.

Underscoring the Unmet Need

SCLC represents approximately 15% of all lung cancer cases, and 70% of these cases are extensive-stage disease, according to Paz-Ares. For these patients, SOC treatment is comprised of induction etoposide, platinum, and an immune checkpoint inhibitor (ICI) in the form of atezolizumab or durvalumab (Imfinzi), followed by maintenance treatment with the same ICI. Although patients respond to induction, many will experience early disease progression and poor survival. Earlier phase trials have shown that when lurbinectedin is paired with ICIs, it is active with favorable tolerability.2-4

Shining a Light on IMforte: Eligibility, Treatment, End Points

The global, open-label, randomized, phase 3 trial enrolled patients with ES-SCLC who had not previously received systemic treatment, who did not have central nervous system metastases, and who have an ECOG performance status of 0 or 1 (n = 660). They received induction treatment with atezolizumab plus carboplatin and etoposide for 4 cycles every 3 weeks.

Patients were screened again and those with an ongoing complete response/partial response or stable disease after induction treatment and an ECOG performance status of 0 or 1 went on to receive maintenance treatment (n = 483). These patients were randomly assigned 1:1 to receive 1200 mg of intravenous atezolizumab every 3 weeks with or without 3.2 mg/m2 of lurbinectedin. Treatment continued until disease progression or intolerable toxicity.

Notably, no crossover was allowed. Patients were stratified by performance status (0 vs 1), lactate dehydrogenase (≤ upper limit of normal [ULN] vs > ULN), liver metastases at baseline induction (yes vs no), and receipt of prophylactic cranial irradiation (yes vs no).

The trial’s primary end points were IRF-PFS and OS, and secondary end points comprised investigator-assessed PFS, objective response rate, duration of response, and safety. Efficacy was evaluated from randomization into the maintenance phase of the design. Investigators assessed safety from day 1 of cycle 1 of treatment.

Additional Efficacy Revelations

Investigator-assessed PFS aligned with what was reported with regard to IRF assessment. The median PFS with lurbinectedin plus atezolizumab was 5.4 months vs 2.7 months with atezolizumab monotherapy (stratified HR, 0.55; 95% CI, 0.45-0.68).

Safety Spotlight

All-cause adverse effects (AEs) were reported in 97.1% of those on the combination arm vs 80.8% of those on the monotherapy arm, with 38.0% and 22.1% of these respective effects being grade 3 or 4. Treatment-related grade 3 or 4 AEs were experienced by more patients in the investigative arm vs those in the control arm, at 25.6% vs 5.8%, respectively.

Twelve patients who received the doublet experienced grade 5 AEs vs 6 patients who received the monotherapy; they were treatment related for 2 patients in the investigative arm and 1 patient in the control arm. More serious AEs occurred with the doublet vs the monotherapy (31.0% vs 17.1%).

AEs led to dose interruption or modification of any drug for 38.0% of those in the investigative arm vs 13.8% of those in the control arm; they led to discontinuation of any drug for 6.2% and 3.3% of patients, respectively.

“The safety profile of lurbinectedin plus atezolizumab was manageable, with mostly low-grade AEs and low treatment discontinuation rates,” Paz-Ares said. “No clinically meaningful increase in immune-related AEs was observed [with the addition of lurbinectedin],” he noted.

The most common AEs experienced by at least 10% of those in the lurbinectedin/atezolizumab and atezolizumab-alone arms were nausea (36.4% vs 4.2%), anemia (31.8% vs 6.7%), fatigue (20.2% vs 7.9%), decreased appetite (16.9% vs 6.7%), decreased platelet count (15.3% vs 2.9%), diarrhea (14.0% vs 7.5%), vomiting (13.6% vs 2.5%), asthenia (12.8% vs 6.3%), thrombocytopenia (12.8% vs 1.7%), decreased neutrophil count (12.8% vs 1.3%), constipation (12.0% vs 6.3%), and neutropenia (10.7% vs 1.7%).

“You see a clear increase for those patients treated with lurbinectedin/atezolizumab, particularly AEs related to the lurbinectedin—nausea, vomiting, diarrhea, asthenia, and also some more myelosuppression,” Paz-Ares noted. “Concerning myelosuppression, that was one of the main concerns, we have seen febrile neutropenia only in 1.7% of cases.”

He added that grade 3 or 4 infections and infestations occurred in 6.6% of those who received the doublet vs 5.0% of those who were given the monotherapy.

Topline Takeaway

“I think it’s important to note that at least in the US, lurbinectedin is FDA approved in the second-line setting, and so, with the results of this study, we would anticipate that it would be moved into the first-line maintenance setting—so, moving it into an earlier line before there is progression on the first-line therapy,” Julie R. Gralow, MD, chief medical officer and executive vice president of ASCO, commented. “The study is important because both PFS and OS were increased. But I would point out, while this is a next step, PFS is still quite low in both arms, and we need to work on additional ways of advancing this even further. So, it is a small next step; it is extending the time that the tumor doesn’t progress and the amount of time that the patients are living, but we need to do more research in ES-SCLC, as well.”

Disclosures: Paz-Ares listed no disclosures.

References

1. Paz-Ares L, Borghaei H, Liu SV, et al. Lurbinectedin (lurbi) + atezolizumab (atezo) as first-line (1L) maintenance treatment (tx) in patients (pts) with extensive-stage small cell lung cancer (ES-SCLC): primary results of the phase 3 IMforte trial. Presented at: 2025 ASCO Annual Meeting; May 30-June 3, 2025; Chicago, IL. Abstract 8006.
2. Calles A, Navarro A, Doger de Speville Uribe BG, et al. Lurbinectedin plus pembrolizumab in relapsed SCLC: the phase I/II LUPAR study. J Thorac Oncol. Published online February 10, 2025. doi:10.1016/j.jtho.2025.02.005
3. Aix SP, 2SMALL (NCT04253145) phase I part: lurbinectedine (LUR) in combination with atezolizumab (ATZ) for second line extensive stage small cell lung cancer (ES-SCLC) patients (pts). J Immunother Cancer. 2021;9. doi:10.1136/jitc-2021-SITC2021.464
4. Aix SP, Navarro A, Garcia MEO, et al. Safety and efficacy of lurbinectedin plus atezolizumab as second-line treatment for advanced small-cell lung cancer: results of the 2SMALL phase 1/2 study (NCT04253145). Presented at: 2025 ASCO Annual Meeting; May 30-June 3, 2025; Chicago, IL. Abstract 8013.

AI in HER2-Low Breast Cancer |

Image Credit: © catalin – stock.adobe.com

Concordance between HER2 immunohistochemistry (IHC) score and the clinical categorization of HER2-expressing tumors was improved among pathologists with the use of artificial intelligence (AI)–assisted biomarker interpretation training, reducing the misclassification of HER2-low and HER2-ultralow cases as HER2-null by 2.24%, according to findings from a study presented during the 2025 ASCO Annual Meeting.

In the study, pathologists were given access to the ComPath Academy training platform, an AI training masterclass that aided in identifying neoplastic cells, total tumor cell count, percentage of each cell class, and a final HER2 score in 20 digital immunohistochemistry (IHC)-stained breast cancer cases. Across 1940 readings from 105 pathologists in 10 countries, the rate of accuracy in pathologists not using AI was 89.1% vs 96.1% in those with AI assistance, while concordance for HER2 scoring was measured at 0.506 for pathologists not using AI vs 0.798 for those using AI. The accuracy in identifying HER2 clinical categories also improved from 90.1% without AI to 95.0% with AI, and concordance among pathologists improved from 0.494 without AI to 0.732 with AI.

Further analysis showed that manual scoring sensitivity was the lowest in cases with no or low levels of HER2 expression, with AI assistance raising the sensitivity across the null, ultralow, and low expression classifications from 54.08% to 88.24%, 50.08% to 93.22%, and 78.64% to 90.35%, respectively. Additionally, HER2-null overscoring without AI assistance was 45.09% vs 11.7% with AI support.

De Brot noted how, while many HER2-low and -ultralow tumors are now targetable, they are often mislabeled as HER2-null, and patients are potentially missing out on access to effective therapies. She pointed to an observed 30% discordance rate among pathologists in scoring HER2-low and -ultralow tumors, underscoring the imperative for advanced training. Reducing misclassification of these cases “potentially enabl[es] more patients to access HER2-directing [antibody-drug conjugate (ADC)] therapies.”

“From a patient perspective as well, what this does is this: with the increasing use and application of anti-HER2-directed therapies, especially in the low and ultralow populations, patients now have access to potentially life-changing medications with this reclassification, demonstrating the fact that there is a potential role for these therapies for these patients,” said Robin Zon, MD, FASCO, FACP, ASCO president, during her expert commentary following the presentation.

In her presentation, De Brot outlined next steps for this research, including rolling out breast HER2 masterclasses to additional countries and pathologists and building a global, unified, international database to map scoring gaps and guide AI training solutions. De Brot also identified a plan to conduct a multicenter implementation study where an AI tool would be embedded in routine diagnostics to measure downstream clinical effects, which would allow investigators to assess changes in treatment allocation and time to therapy in patients with HER2-low and HER2-ultralow breast cancers.

Reference

De Brot, M, Mulder D, Shaaban A, et al. Use of artificial intelligence assistance software for HER2-low and HER2-ultralow IHC interpretation training to improve diagnostic accuracy of pathologists and expand patients’ eligibility for HER2-targeted treatment. Presented at: 2025 ASCO Annual Meeting Pre-Briefing on Regular Abstracts. May 21, 2025. Abstract 1014.

GLP-1 Agonist Use and Cancer Risk

| Image Credit: © Karsyarina

– stock.adobe.com

The use of GLP-1 receptor agonists correlated with a slight decrease in the incidence of obesity-related cancers and was not associated with higher cancer rates compared with the use of DPP-4 inhibitors in patients with diabetes, according to findings from a study presented during a 2025 ASCO Annual Meeting press briefing.1

The study, which included more than 170,000 patients, showed a 7% decrease in obesity-related cancer among patients who used GLP-1 receptor agonists (HR, 0.93; 95% CI, 0.88-0.98; P = .005), who experienced 2501 events vs 2671 events in patients who used DPP-4 inhibitors. Among females, 1754 events were observed with GLP-1 agonist use vs 1898 events with DPP-4 inhibitor use (HR, 0.92; 95% CI, 0.86-0.98; P = .01). Among males, there were 747 events with GLP-1 agonist use and 773 events with DPP-4 inhibitor use (HR, 0.95; 95% CI, 0.86-1.05; P = .29).

There was also an 8% decrease in all-cause mortality with the use of GLP-1 agonists vs DPP-4 inhibitors, with 2783 and 2961 events in each respective group (HR, 0.92; 95% CI, 0.87-0.97; P = .001). Among females, there were 1219 events with GLP-1 agonist use and 1514 events with DPP-4 inhibitor use (HR, 0.80; 95% CI, 0.74-0.86; P < .001). Among males, the respective numbers of events were 1564 with GLP-1 agonist use and 1447 with DPP-4 inhibitor use (HR, 1.04; 95% CI, 0.96-1.11; P = .34).

Investigators assessed 14 types of cancer and found that the protective association between GLP-1 agonists and cancer incidence was driven by cancers of the colon (HR, 0.84; 95% CI, 0.72-0.98; P = .02) and the rectum (HR, 0.72; 95% CI, 0.56-0.93; P = .01). Of note, there was no evidence of adverse association between GLP-1 agonists and pancreatic cancer.

Medullary thyroid cancer could not specifically be assessed in this study due to the small sample size. Although this disease is included in the warning labels for multiple GLP-1–directed medications, there was a lack of association between GLP-1 agonist use and thyroid cancer as a whole in this assessment.

“Findings extend the value of GLP-1[–directed] medicines beyond blood sugar control, weight, heart, and kidney health to potential cancer prevention in adults who are high-risk,” Lucas A. Mavromatis, a research assistant in the Division of Precision Medicine in the Department of Medicine at NYU Grossman School of Medicine, said during the presentation. “The effect size is small, follow-up was short, and assessed medications are primarily weaker, ‘diabetes dose’ formulations; long-term studies are needed to confirm the durability of effect and safety.”

Data were collected between 2013 and 2023 from 43 health system networks and included data from electronic health records and insurance claims. Patients included in the analysis were adults with type 2 diabetes with a body mass index of 30 kg/m2 or more who initiated treatment with either a GLP-1 agonist or a DPP-4 inhibitor for the first time.

Comparison groups included 85,015 pairs, and each GLP-1 starter patient was paired with 1 DPP-4 starter of similar age, sex, race, weight, laboratory values, year of prescription, and medical history. This followed the target trial emulation framework.

Outcomes included the first diagnosis of any 14 obesity-related cancers or death from any cause. Follow-up was from the first prescription until cancer, death, or the last recorded visit, with an average coverage of 3.9 years.

The background for the study was the gap in knowledge of the long-term effect of GLP-1–directed medications on the risk of cancer.

“GLP-1 treatments remain a strong option for people with diabetes and obesity and may favorably affect cancer risk. Decisions should balance benefits, costs, and [adverse] effects in discussion with clinicians,” Mavromatis concluded.

Reference

1. Mavromatis LA, Surapaneni A, Mehta S, et al. Glucagon-like peptide-1 receptor agonists and incidence of obesity-related cancer in adults with diabetes: a target-trial emulation study. Presented at: 2025 ASCO Annual Meeting; May 30-June 3, 2025; Chicago, IL. Abstract 10507.

Artificial Intelligence
in Cancer Treatment
| Image Credit: © Crystal light
– stock.adobe.com

As artificial intelligence (AI) seeps more and more into the daily lives of professionals in every industry, cancer care has proven to be no different. In fact, AI-based approaches are already having a dramatic effect on multiple areas of oncology, including treatment selection, the lowering of administrative burden, and drug development.

“One of the most timely and pressing [concerns] in modern oncology is how AI is shaping the clinical care environment, and whether it is truly improving efficiency, communication, and outcomes,” Enrique Velazquez Villarreal, MD, PhD, MPH, MS, assistant professor, Department of Integrative Translational Sciences at City of Hope in Duarte, California, said in an interview with OncLive®. “Real-world applications of AI are already entering the clinic [in forms such as] ambient listening tools, patient messaging chat bots, and inbox management systems, all of which aim to reduce administrative burden and enhance patient-[clinician] communication. [We are] at the point when we are starting to use the available models, [which] is the first step in clinicians exploring the [existing] options.”

AI Is Already Making a Splash in the Clinic

AI-based approaches, including natural language processing and machine learning, are already having a dramatic effect on the ways that patients with cancer can be diagnosed and treated, as well as showing potential in drug development.

One example is a machine learning model developed by a team of investigators from the European Society for Blood and Marrow Transplantation (EBMT) which can predict overall survival for patients with myelofibrosis following allogeneic hematopoietic cell transplantation (allo-HCT).1 The model accounts for patient age,comorbidity index, performance status, blood blasts, hemoglobin, leukocytes, platelets, donor type, conditioning intensity, and graft-vs-host disease prophylaxis.2 In March 2025, the open-access tool, which features an interactive web-based calculator that visualizes the risk score based on inputs of the 10 variables, was launched online and it is free to use.1

Data from 5,183 patients across 288 EBMT-registered centers were used to train the machine learning model.2 Findings from a study published in Blood demonstrated that the model identified 25% of patients as being at high risk for a poor outcome with allo-HCT, defined as 1-year non-relapse and overall mortality rates of approximately 35% and 40%, respectively. Comparatively, traditional methods using Cox-derived and Center for International Blood and Marrow Transplant Research scores assigned patients to the high-risk group at respective rates of 10.1% and 8.2%.

“We currently have good [transplant risk projection] scoring systems, but you [also] have to consider the toxicity and the results of transplant, and we don’t have very good tools at this time [for that],” Juan Carlos Hernández-Boluda, MD, PhD, associated physician, Hematology and Medical Oncology Service, Hospital Clínico of Valencia, Spain said in an interview with OncLive. “With this [machine learning] tool, we will have information on the disease and transplantation risk, and we can share this information with the patient. We also have a web application that is visual, and we can share this image with the patient if they are interested. [Using this tool], we can make a more informed decision, which is critical because transplantation is a curative treatment, but also it poses a lot of danger [in terms of] the life of the patient. Combining what we had before with this tool will help us to make that decision.”

AI has also already shown its value in reducing administrative burdens by mining electronic health records (EHRs) for valuable patient data much faster and more accurately compared with human capabilities.3 In a study published in Nature Communications, investigators developed a natural language processing model that can extract structured cancer outcome variables from EHRs for creation of multi-site clinico-genomic datasets.

The ‘teacher’ portion of the model was trained using EHR data from Dana-Farber Cancer Institute (DFCI) to label imaging reports and discharge summaries. The ‘student’ portion of the model then used this information to predict the labels that were assigned by the teacher component and sent the information to Memorial Sloan Kettering Cancer Center (MSKCC) for evaluation.

The DFCI cohort included 37,274 annotated imaging reports from 3,213 patients and 39,191 annotated oncologist notes from 3,588 patients, across 7 cancer types. The MSKCC cohort included 24,472 annotated imaging reports from 2,672 patients and 40,701 annotated oncologist notes from 3,617 patients, across 5 cancer types. Findings from the study showed that the model’s approach was feasible in terms of extracting clinical variables from the EHR across cancer centers. Importantly, the model did not expose protected health information to investigators during its training. The study authors concluded that this approach could significantly expand the scale of clinical datasets for use in precision oncology research.

Additionally, AI-based approaches have shown great potential in terms of reducing research costs and increasing the speed of drug development.4 For example, AI has shown the ability to predict drug-target interaction performance, achieve target prediction accuracy of approximately 90% on over 2,000 small molecules, and the capability of predicting the druggability of anticancer drug targets. It can also identify drugs that could be repurposed based on drug-target interaction and assist in the accurate prediction of reactions to anticancer agents.

Addressing the Current Limitations of AI in Cancer Care

Despite the great promise that AI has already displayed in cancer care, there are concerns among experts that some of the limitations of the technology still need to be improved before it is truly ready for universal adoption. One issue that has concerned investigators is the presence of biases that can be introduced during the training of AI models, rendering them ineffective for some groups of patients.

“We don’t want to introduce biases or errors, particularly if they are going to impact patient care, so it’s a work in progress,” Mack Roach III, MD, professor of radiation oncology, medical oncology and urology at the University of California San Francisco, explained in an interview with OncLive. “There have been missteps along the way. For example, a model can be created in one population that’s made up of individuals with fair skin, then you try to to look at the retina to see [whether] a patient is at risk for retinal disease. If it is applied in a population where individuals have more pigmentation [compared with those it was trained with], the AI model isn’t going to work effectively.”

In a study published in JCO Clinical Cancer Informatics, Roach and his coauthors developed a novel multimodal AI deep learning system that incorporated digital histopathology and clinical data.5 The study authors aimed to assess whether the tool could be applied across risk-stratified African American patients with prostate cancer without algorithmic bias compared with non–African American patients.

The study included data from the randomized phase 3 NRG/RTOG 9202, 9408 (NCT00002597), 9413 (NCT00769548), 9910 (NCT00005044), and 0126 (NCT00033631) studies, in which all patients were treated with definitive external radiotherapy with or without androgen deprivation therapy. The study included high-quality digital histopathology data from 5,708 patients from the 5 studies; 948 were African American and 4,731 were non–African American.

Findings from the study showed that the model performed similarly in terms of showing a strong prognostic signal for distant metastasis in both the African American (subdistribution HR, 1.2; 95% CI, 1.0-1.3; P = .007) and non–African American (subdistribution HR, 1.4; 95% CI, 1.3-1.5; P < .001) groups. Additionally, the multimodal AI model for prostate cancer–specific mortality displayed a strong prognostic signal in both groups, with subdistribution HRs of 1.3 (95% CI, 1.1-1.5; P = .001) and 1.5 (95% CI, 1.4-1.6; P < .001), respectively.

“These men were treated on prospective phase 3 randomized trials, so we feel that this is the most reliable information that’s available,” Roach said. “[Our findings] demonstrate algorithms can be created based on AI that improve our ability to predict prognosis and guide treatment without introducing bias.”

“AI algorithms are now being applied in terms of deciding [which patients] should receive hormone therapy with radiation and who should not, as well as who should receive long-term vs short-term hormone therapy. Having established as a baseline that the algorithm did not introduce bias, we would deduce that if we’re deciding how long a patient should get hormone therapy and whether they should get hormone therapy or not, that we don’t have to worry about introducing bias. [These decisions] have a major impact not only on the patient’s survival but also quality of life, because patients that don’t need [therapy] won’t be receiving it. The downstream effects in terms of what to do for an individual patient are a real benefit.”

Investigators Anticipate Upcoming AI Research

During the upcoming 2025 ASCO Annual Meeting, several planned sessions will explore the role of AI in areas including drug development, community practice, disease screening, disease assessment, imaging, and clinical decision support. Abstracts will also be presented during oral sessions on a multimodal AI model used to identify benefit from second-generation androgen receptor inhibitors in patients with high-risk nonmetastatic prostate cancer and an AI-based platform designed for dynamic visualization of the National Comprehensive Cancer Network guidelines for patients with breast cancer.

“The question [regarding the future of AI in cancer care] should almost be, ‘Where will it not be useful?’” Roach said. “We need to have good data from studies available, then patients will be able to independently get a second opinion from both their doctors [and] the computer. The only risk there is that we need to provide accurate information; we need to make sure that we don’t have any biases. AI may also [prove to] be useful in evaluating the quality of publications [by] identifying false information included in studies. [It will be able to] identify fraudulent manuscripts [as well as] evaluate the quality of conclusions, providing transparency. There is tremendous potential [overall].”

References

1. Machine learning program enhances transplant risk assessment in myelofibrosis patients better than current models. News release. American Society of Hematology. March 27, 2025. Accessed May 20, 2025. https://www.hematology.org/newsroom/press-releases/2025/machine-learning-program-enhances-transplant-risk-assessment-in-myelofibrosis-patients
2. Hernandez-Boluda JC, Mosquera Orgueira A, Gras L, et al. Use of machine learning techniques to predict poor survival after hematopoietic cell transplantation for myelofibrosis. Blood. Published online March 27, 2025. doi:10.1182/blood.2024027287
3. Kehl KL, Jee J, Pichotta K, et al. Shareable artificial intelligence to extract cancer outcomes from electronic health records for precision oncology research. Nat Commun. 2024;15(1):9787. doi:10.1038/s41467-024-54071-x
4. Wang L, Song Y, Wang H, et al. Advances of artificial intelligence in anti-cancer drug design: a review of the past decade. Pharmaceuticals (Basel). 2023;16(2):253.doi:10.3390/ph16020253
5. Roach M 3rd, Zhang J, Mohamad O, et al. Assessing algorithmic fairness with a multimodal artificial intelligence model in men of African and non-African origin on NRG oncology prostate cancer phase III trials. JCO Clin Cancer Inform. Published online May 9, 2025. doi:10.1200/CCI-24-00284