The research was led by Jia Wu, Ph.D., associate professor of Imaging Physics and of Thoracic/Head and Neck Medical Oncology at MD Anderson and an affiliate member of the institution’s Institute for Data Science in Oncology. Co-senior authors include Ajay Sheshadri, M.D., associate professor of Pulmonary Medicine, and Mehmet Altan, M.D., associate professor of Thoracic/Head and Neck Medical Oncology. Their team set out to solve a problem that has long frustrated oncologists: immune checkpoint inhibitors have transformed outcomes in lung cancer, but the same drugs that unleash T cells against tumors can also trigger the immune system to attack healthy lung tissue, producing pneumonitis that is difficult to anticipate before symptoms appear.

“Pneumonitis remains one of the most challenging complications of immunotherapy because it can be difficult to predict before symptoms appear,” Wu said. “Our model was able to identify signals associated with future risk using information that already exists in routine CT scans.” That distinction matters clinically, because once symptomatic pneumonitis develops, patients often require high-dose corticosteroids or other immunosuppressive therapy, and in severe cases the drugs that were controlling their cancer must be stopped entirely. A method for identifying high-risk patients before treatment could allow oncologists to schedule closer monitoring, order earlier imaging follow-up, or enroll vulnerable patients in prevention trials.

The model, named the Checkpoint-Inhibitor Pneumonitis Hazard EstimatoR, or CIPHER, belongs to a class of systems known as AI foundation models. Rather than being trained narrowly on labeled examples of pneumonitis, the model was first trained on an enormous volume of chest CT data, more than 590,000 individual image slices drawn from 2,500 patients with lung cancer. During this phase it learned to recognize the underlying structural patterns of lung tissue: the fine meshwork of airways and vessels, the texture of healthy parenchyma, and the visual signatures of scarring, emphysema, and inflammation. This strategy, sometimes called self-supervised or transfer learning, lets the model build a rich internal representation of normal and abnormal lung anatomy before it is ever asked to make a clinical prediction.

Only after learning lung anatomy did the researchers ask whether the patterns CIPHER had absorbed could distinguish patients who later developed pneumonitis from those who did not. The team tested the model on pretreatment CT scans from 347 patients with non-small cell lung cancer treated at UT MD Anderson, and then validated it against an independent external dataset. In both cohorts, CIPHER achieved an area under the curve, or AUC, of approximately 0.83, a statistical measure in which 0.5 represents chance-level guessing and 1.0 represents perfect discrimination. An AUC of 0.83 indicates strong predictive performance, and critically, the model outperformed both conventional models built on clinical risk factors and radiomics approaches, which extract hand-designed quantitative features from images.

Robustness across different conditions is often the stumbling block for medical AI, and it is where CIPHER delivered some of its most persuasive results. The external validation cohort differed from the original in patient population, in the types of CT scanners used, and in imaging protocols, variables that frequently degrade the performance of image-based models when they are deployed outside their training environment. CIPHER nonetheless maintained its level of accuracy, suggesting the model captured genuine biology rather than scanner-specific artifacts or population quirks. Patients the model classified as high-risk also tended to develop pneumonitis sooner after starting immunotherapy, an indication that the model may be measuring a continuous degree of lung vulnerability rather than simply sorting patients into binary categories.

The team further tested whether CIPHER’s predictions stood up when adjusted for known clinical variables. Even after accounting for age, smoking history, tumor histology, and prior radiation therapy to the chest, factors that independently influence pneumonitis risk, the model’s predictions remained statistically significant. Prior thoracic radiation in particular is recognized as a major contributor to treatment-induced lung injury, so a model that adds predictive value on top of these established risk factors is capturing something that clinicians currently cannot see. “What makes this approach particularly interesting is that it was not designed to look for pneumonitis itself,” Wu said. “Instead, the model learned patterns within lung tissue and identified subtle abnormalities associated with future risk. That suggests routine imaging may contain much more information about treatment toxicity than we previously recognized.”

The findings arrive at a moment when immune checkpoint inhibitors have become a cornerstone of lung cancer therapy. These drugs work by releasing molecular brakes, such as the PD-1 and CTLA-4 pathways, that normally restrain T cell activity, allowing immune cells to recognize and attack tumor cells. The same mechanism, however, can produce immune-related adverse events in nearly any organ, with pneumonitis among the most feared because of its potential severity and because it can be difficult to distinguish from infection, tumor progression, or radiation injury. Current risk assessment relies on subjective radiologic review and broad clinical factors that fail to identify most of the patients who will ultimately be affected, leaving clinicians to detect pneumonitis only after cough, shortness of breath, or infiltrates on follow-up imaging appear.

The MD Anderson researchers caution that several steps remain before CIPHER can enter routine clinical use. Additional prospective studies involving larger and more diverse patient populations will be needed to determine whether the approach can be integrated into clinical workflows, where it would ideally run automatically on scans already being obtained for staging and treatment planning. The team also plans to evaluate whether the model performs similarly in other cancer types treated with immunotherapy, since checkpoint inhibitors are now used across melanoma, kidney, bladder, and many other malignancies, each with different baseline lung characteristics and treatment contexts.

Future work may also explore whether combining CIPHER’s imaging-based risk scores with other biomarkers, such as blood-based inflammatory markers or genetic signatures, can sharpen predictions further, and whether similar foundation models can forecast additional immunotherapy-related toxicities affecting other organs. If validated, such tools could identify candidates for prevention studies, guide the intensity of monitoring during treatment, and deepen scientific understanding of how immune-related side effects develop in the first place. The research was supported by the National Institutes of Health, the Cancer Prevention and Research Institute of Texas, and institutional funding from UT MD Anderson, and the study was published in the Journal for ImmunoTherapy of Cancer.

Subject of Research: An AI foundation model that predicts immune checkpoint inhibitor-induced pneumonitis risk in lung cancer patients from routine pretreatment CT scans.

Article Title: AI model uses routine imaging to identify patients at risk for serious treatment-induced lung inflammation

Article References: AI model uses routine imaging to identify patients at risk for serious treatment-induced lung inflammation. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: artificial intelligence, immunotherapy, pneumonitis, lung cancer, CT imaging, deep learning, checkpoint inhibitors, MD Anderson, risk prediction, radiomics, non-small cell lung cancer, medical imaging

The review, led by Lotte M.G. Hulskotte and colleagues in the Netherlands, systematically searched PubMed and Embase for evidence on drug–drug and food–drug interactions and pharmacodynamic adverse effects of all thirty SMIs approved by the US Food and Drug Administration and the European Medicines Agency for non-small cell lung cancer, from long-standing agents such as erlotinib and crizotinib to newcomers including zongertinib, sevabertinib and sunvozertinib. The searches, initiated in October 2025 and finalised in July 2026, were supplemented with regulatory documents from both agencies, and the authors applied a pragmatic hierarchy of evidence ranking pharmacokinetic studies and meta-analyses above phase 1/2 trials, observational data, label information and expert opinion.

At the heart of the pharmacokinetic problem lies chemistry. Most SMIs are weakly basic, lipophilic molecules whose absorption depends delicately on the acidity of the stomach. When acid-reducing agents such as proton pump inhibitors raise gastric pH, the drugs shift toward a less soluble, non-ionised form that dissolves poorly, cutting systemic exposure. The magnitude is striking: selpercatinib exposure fell by roughly 70 percent when omeprazole was given in the fasted state, and sotorasib lost 57 percent of its exposure with omeprazole even when taken with food. Erlotinib, co-administered with esomeprazole, lost nearly half its area under the curve. Bioequivalence standards allow exposure to vary only within 80 to 125 percent of the reference value, so deviations of this size translate directly into subtherapeutic concentrations, lost efficacy and, ultimately, disease progression.

Liver enzymes provide the second major axis of interaction. Most SMIs are metabolised primarily by cytochrome P450 iso-enzymes, above all CYP3A, so potent inducers such as rifampicin can collapse drug levels while inhibitors such as itraconazole can push them to toxic heights. Rifampicin cut adagrasib exposure by 95 percent, while itraconazole raised repotrectinib exposure nearly sevenfold. Compounding the problem, several agents are themselves auto-inducers or auto-inhibitors of CYP enzymes — dabrafenib, encorafenib, lorlatinib and osimertinib induce their own clearance, while adagrasib, ceritinib and taletrectinib inhibit it — meaning that single-dose interaction studies may not predict steady-state behaviour. Transporters such as P-glycoprotein and breast cancer resistance protein add further complexity, although their intestinal contribution is generally considered minor.

Food, by contrast, can be an ally. High-fat meals delay gastric emptying, increase bile flow and create a lipophilic environment that enhances solubilisation of many of these drugs. Alectinib is the poster child: a high-fat meal raised its systemic exposure by roughly 230 to 250 percent, and a clinical crossover study showed that 35 percent of patients failed to reach the alectinib exposure threshold when taking the drug with low-fat yoghurt, compared with only 5 percent eating a continental breakfast or lunch. This very property has been exploited therapeutically — ceritinib’s approved dose fell from 750 mg fasted to 450 mg with food after feeding studies showed comparable exposure with less gastrointestinal toxicity, and the FDA has required a post-marketing evaluation of taletrectinib dosing with food for the same reason. Food also reduced inter-individual variability for several agents, improving predictability of therapeutic exposure.

The interplay between food and acid suppression, however, is drug-specific and sometimes counterintuitive. Selpercatinib’s PPI-induced exposure loss is largely rescued by co-administration with a meal. Sotorasib behaves in the opposite way, losing even more exposure when food and a PPI are combined. For erlotinib, an acidic glass of cola partially reversed the damage done by esomeprazole, restoring roughly 40 percent of lost exposure. The authors conclude that when PPI use is unavoidable, clinicians should weigh the individual pharmacokinetics of each SMI, considering food co-administration, acidic beverages or staggered dosing as mitigation strategies.

Turning to the drugs’ intrinsic toxicities, the review paints a landscape where class effects mask considerable agent-to-agent variation. Corrected QT interval prolongation, which can degenerate into dangerous arrhythmias, affects the entire class but is most pronounced with adagrasib, ceritinib, crizotinib, encorafenib, osimertinib and taletrectinib, with median increases exceeding ten milliseconds and the highest rates of clinically significant prolongation seen with adagrasib, encorafenib, taletrectinib, entrectinib and selpercatinib. Because the effect is largely exposure-dependent, any interaction that raises drug levels also raises arrhythmia risk. Others, including alectinib, gefitinib, capmatinib and the newer HER2 inhibitors, showed no clinically meaningful QTc changes. Management centres on electrocardiographic monitoring, correction of concomitant QT-prolonging factors, and dose interruption or discontinuation when the QTc exceeds 500 milliseconds.

Pneumonitis, the non-infectious inflammation of lung tissue that can prove fatal, occurs with most SMIs but clusters around particular agents. Brigatinib carried the highest trial incidence at 9 percent of all-grade events, with an unusual early onset within the first week — the rationale for its mandatory one-week 90 mg lead-in dose before escalation. Pralsetinib affected about 12 percent of patients including one fatal case, while adagrasib, capmatinib, ensartinib and sunvozertinib each produced events in 5 to 6 percent. Strikingly, ethnicity emerged as an independent risk factor: all-grade interstitial lung disease was reported in up to 15 percent of Japanese patients on gefitinib and 17 percent on osimertinib. Rechallenge after pneumonitis remains fraught — one real-world study found a 50 percent twelve-month recurrence rate when patients were rechallenged with osimertinib itself, versus 15 percent with alternative EGFR inhibitors.

Among the quieter but increasingly recognised toxicities is weight gain. NTRK inhibitors cause weight gain in 53 percent of patients, likely through on-target inhibition of TRKB, a hypothalamic receptor governing appetite, with entrectinib reaching roughly 66 percent incidence. Among ALK inhibitors, lorlatinib produced all-grade weight gain in up to 81 percent of patients, with grade 3 events in up to 23 percent and a median gain of 4.5 kilograms in a prospective study; alectinib added an average 9 centimetres of waist circumference over a year. The irony is acute for alectinib, whose absorption depends on high-fat meals. Management spans lifestyle counselling, systematic weight monitoring and, where needed, GLP-1 receptor agonists — though even here caution is warranted, as semaglutide was recently shown to reduce alectinib exposure by 32 percent.

Further distinguishing the class, EGFR inhibitors uniquely damage the ocular surface — keratitis in 1 to 3 percent of patients and conjunctivitis up to 24 percent with dacomitinib — because EGFR sustains corneal epithelium and tear production, while BRAF/MEK combinations bring uveitis and retinal pigment epithelial detachment, the latter affecting roughly 30 percent of patients on encorafenib plus binimetinib. Central neurotoxicity, including cognitive disorders in up to a third of patients on capmatinib, entrectinib, lorlatinib and repotrectinib, peripheral neuropathy in up to 44 percent on lorlatinib, dabrafenib-driven pyrexia in 39 percent of combination-treated patients, severe stomatitis with second-generation EGFR inhibitors, and hepatotoxicity necessitating intra-class switches — erlotinib after gefitinib injury, adagrasib after sotorasib injury — complete the picture. The authors’ overarching message is that safe prescribing of these transformative drugs demands equal attention to what patients eat, what else they swallow, and which specific molecule sits in the capsule: pharmacokinetic–pharmacodynamic associations remain under-studied, and filling that gap could enable personalised dosing that maximises efficacy while sparing patients avoidable harm.

Subject of Research: Pharmacokinetic and pharmacodynamic considerations for prescribing small molecule kinase inhibitors in non-small cell lung cancer

Article Title: Pharmacological considerations for prescribing of small molecule inhibitors in patients with non-small cell lung cancer

Article References: Hulskotte, L. M., Veerman, G. M., Lanser, D. A., Reyners, A. K., van Schaik, R. H., Dingemans, A.-M. C., Taxis, K., Mathijssen, R. H., & Jansman, F. G. (2026). Pharmacological considerations for prescribing of small molecule inhibitors in patients with non-small cell lung cancer. eClinicalMedicine, 100, Article 104199. https://doi.org/10.1016/j.eclinm.2026.104199

Image Credits: AI Generated

DOI: 10.1016/j.eclinm.2026.104199

Keywords: non-small cell lung cancer, small molecule inhibitors, drug-drug interactions, pharmacokinetics, CYP3A, proton pump inhibitors, food-drug interactions, QTc prolongation, pneumonitis, weight gain, ocular toxicity, targeted therapy