Bibliographic record
Abstract
A large retrospective study of a prediction tool for metastatic risk in early melanoma proved “robust, precise, and applicable” across multiple international populations, new research showed. Investigators in the collaborative, multinational study analyzed the tool's ability to calculate the risk of cancer spread in 15,731 patients who underwent sentinel lymph node biopsy for cutaneous melanoma. The data came from newly diagnosed melanoma patients in four continents, including cancer centers in both the U.S. and the U.K. between July 2021 and December 2023. All of the patients had three mandatory risk factors: age at diagnosis, Breslow thickness of the primary melanoma, and melanoma subtype. Also available, but only for some, were ulceration status, lymphovascular invasion, and the mitotic rate for how rapidly tumor cells divided. Results of the analysis, published recently in JAMA Dermatology, found that the prediction tool, initially developed and validated by the Melanoma Institute Australia (MIA) in 2020, performed as well overall as the original model (2025; https://doi.org/10.1001/jamadermatol.2025.0318). Moreover, given the much larger patient dataset in the current study, it did so with better precision—a “tighter fit.” “These findings should provide users globally with greater confidence when using the tool in day-to-day clinical practice,” the investigators said. The MIA risk-prediction tool belongs to a crowded field of about 20 risk-prediction models for melanoma, which oncologists often use to stage patients' disease and treatment through sentinel node biopsy. But, in a systematic review and meta-analysis done by researchers at the University of Calgary, Alberta, earlier this year, the MIA model and that of Memorial Sloan Kettering Cancer Center (MSKCC) emerged as the most well-validated of these tests with “strong and comparative discriminative performance.” Both risk models already help inform doctors' decision-making process for performing sentinel node biopsy for suspicious lesions under guidelines established by the National Comprehensive Cancer Network (NCCN). Updated recently, the NCCN guidelines now recommend that doctors use broader criteria than they have in the past—beyond Breslow thickness and ulceration status alone—in deciding whether to do one of these procedures. In the new analysis, investigators saw the best predictive results when patients had not only the three mandatory parameters for risk, but all three optional risk factors as well. Although only 20 percent of patients had these six data elements, “the analysis supports that the model performs better when all the elements are included,” said Jeffrey Gershenwald, MD, Professor in the Department of Surgical Oncology, Division of Surgery, at The University of Texas MD Anderson Cancer Center. Moreover, with the inclusion of additional data, the confidence intervals associated with risk-point estimates were smaller, he said, suggesting greater confidence in the model as a tool for informing clinical discussions with patients. Current guidelines generally recommend sentinel node biopsy for patients whose risk for metastatic disease is considered at least 10 percent and to offer and discuss these procedures when a patient's risk is at least 5 percent, according to Gershenwald. But for many patients diagnosed with a surface lesion, such as melanoma in situ, involving only the epidermis, for which the risk of synchronous regional metastasis is essentially non-existent, or a very early invasive melanoma with limited dermal involvement, he said “this discussion and this tool would not generally be applicable.” Once it involves more of the dermis and has other worrisome, high-risk features, however, “that's when the use of the templated pathology report can be used with this tool to assess and discuss the risk with a patient.” Sentinel node biopsy itself carries a small risk of complications, estimated at around 5 percent. “Most prominently, you might see a fluid pocket, which sometimes persists,” leading to a seroma or infection, Gershenwald said. “There's always a balancing act between the healing process and the space where the lymph node has been removed.” In the future, head-to-head comparisons of these risk-prediction tools are deemed unlikely. Researchers at both the Memorial Sloan Kettering Cancer Center and the Melbourne Institute Australia have done extensive comparisons in the past, according to Alexander Varey, PhD, Clinical Associate Professor, Director of Surgical Research at the University of Sydney, and principal investigator of the JAMA study. “In both our dataset and [that of] the MD Anderson Cancer Center dataset [in 2020], we found consistently better performance with the MIA tool,” Varey said. Some studies have found the two leading prediction models similar, he acknowledged, but “we found the MSKCC tool underestimated risks, particularly in the lower risk range, where the tools are more important in decision making.” As for adding gene expression profiling for further refinement, that remains to be seen. Depending on whether adding genetic markers improves the predictive power of sentinel node biopsy, the study authors say it will have to be large enough to justify the anticipated hike in cost. Susan Jenks is a contributing writer.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".