A groundbreaking new study supports the potential of electronic health records as a tool for predicting the risk for gastric cancer. Using a predictive model that analyzes everyday clinical data, the investigators can uncover individuals at high risk for cancer who may need early endoscopic screening, potentially changing how gastroenterologists practice cancer prevention.
Early detection plays a key role in enhancing outcomes in patients with gastric cancer, but traditional screening methods can be invasive and require significant resources. The findings suggest that clinicians can use existing patient data from the electronic health record (EHR), such as lab results and clinical messages, to predict cancer risk in a more efficient and noninvasive way.
“It’s incredibly powerful that we can use real-world data from everyday clinical practices to predict something as serious as gastric cancer,” lead investigator Michelle Kang Kim, MD, PhD, the chair of the Department of Gastroenterology, Hepatology and Nutrition at Cleveland Clinic in Ohio told Gastroenterology & Endoscopy News. Dr. Kim emphasized the importance of using the EHR to improve patient care and drive forward scientific research. “Our ultimate goal is to expand this predictive model to cover other common diseases as well. By doing so, we can identify high-risk patients early and intervene before the disease progresses.”
Dr. Kim and her co-investigators included 614 patients between 40 and 80 years of age who were diagnosed with noncardia gastric cancer (NCGC) and were treated at a large tertiary medical center between 2010 and 2021 (Gastro Hep Advances 2024;3:910-916). They randomly selected controls without a diagnosis of NCGC in a 1:10 ratio of cases to controls.
The investigators assessed how accurately patients are classified as having or not having NCGC using the area under the receiver operating characteristic curve and the 0.632 estimator. To estimate the probability of NCGC, they performed multiple imputation by chained equations for missing data followed by logistic regression on imputed data sets. To identify an optimal threshold to be used to select patients for screening, they maximized the sensitivity for a fixed value of specificity based on the average predicted probability of gastric cancer across the multiple imputed data sets. Positive predictive value (PPV) also was calculated for different threshold points.
The 0.632 estimator value was 0.731, indicating robust model performance. The probability of NCGC was higher with increasing age (odds ratio [OR], 1.16; 95% CI, 1.04-1.3), male sex (OR, 1.97; 95% CI, 1.64-2.36), Black (OR 3.07; 95% CI, 2.46-3.83) or Asian race (OR, 4.39; 95% CI, 2.60-7.42), tobacco use (OR, 1.61; 95% CI, 1.34-1.94), anemia (OR, 1.35; 95% CI, 1.09-1.68), and pernicious anemia (OR, 6.12; 95% CI, 3.42-10.95). In contrast, the investigators found that patients with liver disease (OR, 0.53; 95% CI, 0.33-0.85), hypertension (OR, 0.7; 95% CI, 0.56-0.87) or hypercholesterolemia (OR, 0.38; 95% CI, 0.30-0.47) had less risk for NCGC.
Different thresholds of the model achieved varying sensitivity and specificity values, but the most specific model achieved a PPV of 1%, which Dr. Kim and her team noted is “approaching the desired value for screening tests.”
The findings are crucial for practicing gastroenterologists, according to Dr. Kim. Predicting gastric cancer risk using easily accessible data could transform how screenings are performed, leading to more precise and effective methods. “The integration of data science into daily practice is not just an advancement in technology but a significant step forward in personalized medicine,” Dr. Kim noted.
This predictive model shows potential as a valuable research tool. By analyzing patient data, healthcare providers can uncover hidden trends and patterns that can greatly affect the field of gastroenterology and other medical fields.
Dr. Kim acknowledged that while improvements are needed, the current results are a crucial first step. “We are at the forefront of a new way of thinking about patient data and disease prediction. The fact that we can even achieve a 1% PPV at this stage is a testament to the potential of this approach.”
Rashmi Advani, MD, an assistant professor of medicine at the Icahn School of Medicine at Mount Sinai, in New York City, agreed with Dr. Kim on the study’s significance. “With the global rise in gastric cancer cases, developing reliable risk prediction models is essential. This study provides the foundational framework needed to identify asymptomatic patients who could benefit from early endoscopic screening based on their risk factors,” she explained.
Dr. Advani also brought attention to the wide implications of this research. “It’s not just about gastric cancer. This approach can be applied to other conditions, allowing us to use data more effectively to improve patient outcomes. By identifying high-risk individuals early, we can take proactive steps to prevent disease progression.”
Dr. Kim is optimistic about the impact of their research on a larger scale. “We’re not just looking at gastric cancer. We’re aiming to create a tool that can be used across various diseases. This is about harnessing the power of data to make healthcare more predictive, personalized and preventive.”
—Morgan Gress
This article is from the March 2025 print issue.
link