The clinical problem the researchers tackled is deceptively simple to state but stubbornly difficult to solve. When a woman is diagnosed with invasive breast cancer, determining whether the disease has reached the regional lymph nodes is central to staging, adjuvant therapy planning, and surgical strategy. The current standard for patients whose nodes feel normal and look normal on ultrasound is the sentinel lymph node biopsy, a technique that identifies and removes the first draining node or nodes so they can be examined under the microscope. Yet even this minimally invasive procedure carries real costs: lymphedema, seroma formation, numbness, restricted shoulder mobility, and the anxiety of a second operation when metastases are found. For a substantial fraction of patients, that biopsy ultimately confirms no nodal involvement at all, meaning the procedure was, from a purely diagnostic standpoint, avoidable.

The Madrid cohort comprised 540 consecutive patients with invasive HR+/HER2- breast cancer treated between January 2019 and June 2024, all of whom underwent breast surgery with formal axillary staging. This tumor subtype is the single largest biological category of breast cancer, driven by estrogen or progesterone signaling rather than HER2 amplification, and typically managed with endocrine therapy, chemotherapy in selected cases, and surgery. Because these tumors often behave in a more indolent fashion than triple-negative or HER2-positive disease, the question of which patients can safely forgo parts of the axillary workup is particularly pressing. In this cohort, 173 patients, or 32.0 percent, turned out to have nodal disease despite being classified as clinically node-negative before surgery, a figure that underscores how much microscopic spread evades conventional preoperative assessment.

To build the prediction model, the team assembled a rich set of preoperative variables spanning clinical examination, imaging, and core needle biopsy pathology: tumor size, multifocality or multicentricity, histological type and grade, lymphovascular invasion, perineural invasion, hormone receptor and Ki-67 values, and the findings of axillary ultrasound. Rather than relying on univariate associations, which can be confounded by correlations among variables, the investigators used multivariable logistic regression with LASSO regularization. LASSO, short for least absolute shrinkage and selection operator, is a statistical technique that penalizes model complexity, shrinking weak coefficients toward zero and effectively eliminating predictors that add little independent information. This approach guards against overfitting, the failure mode in which a model memorizes the idiosyncrasies of its development dataset and then performs poorly on new patients.

The variables that survived LASSO selection tell a coherent biological story. By far the strongest predictor was a suspicious axillary ultrasound, which carried an odds ratio of 16.42, meaning patients with abnormal-appearing nodes on imaging had more than a sixteenfold increase in the odds of harboring metastases compared with those whose ultrasound was unremarkable. Lymphovascular invasion on the preoperative biopsy, an indication that tumor cells have already entered lymphatic or blood vessels within the primary tumor, roughly doubled the odds at an odds ratio of 2.61. Each additional millimeter of tumor size increased the odds by about 4 percent, and multifocal or multicentric disease, where multiple tumor foci are present in the breast, raised the odds by 88 percent. Histological grade was retained in the final model for reasons of clinical relevance even though its statistical contribution was more modest.

Performance metrics matter as much as the predictors themselves, and here the nomogram delivered solid numbers. Discrimination, the model’s ability to separate patients who have nodal disease from those who do not, was quantified by the area under the receiver operating characteristic curve, which reached 0.83 with a 95 percent confidence interval of 0.79 to 0.87. In practical terms, an AUC of 0.83 means that if the tool were given one patient with nodal involvement and one without, it would assign the higher risk score to the correct patient roughly 83 percent of the time, a level generally regarded as good for a clinical prediction model built on routine preoperative data. Calibration, which asks whether predicted probabilities match observed frequencies across the risk spectrum, was equally reassuring: the expected-to-observed ratio was 1.002, essentially perfect, and the Brier score, a composite measure combining discrimination and calibration, was 0.135, comfortably below the 0.25 that would correspond to uninformative guessing.

A nomogram translates these regression coefficients into a visual scoring system that clinicians can use at the bedside. Each predictor is assigned a point value on a scale, the points for all variables are summed, and the total maps onto a predicted probability of nodal involvement. The Madrid team went further by calculating diagnostic performance across multiple probability thresholds, reporting positive and negative predictive values at each cut-off. This threshold analysis is what makes the tool actionable rather than merely descriptive: a hospital could, for example, define a low-risk band below which sentinel biopsy might be omitted or deferred, or a high-risk band above which more intensive axillary evaluation or direct discussion of nodal therapy could be considered. The authors explicitly frame the nomogram as a support for individualized risk stratification within the broader movement toward axillary surgical de-escalation.

That de-escalation movement has been reshaping breast surgery for two decades. Following landmark trials such as ACOSOG Z0011, which showed that many women with limited sentinel node metastases who undergo breast-conserving therapy do not benefit from completion axillary lymph node dissection, guidelines have progressively relaxed the requirement for radical nodal surgery in selected patients. More recently, trials of omitting sentinel biopsy altogether in older, low-risk patients have gained traction. The HR+/HER2- subtype is a natural focus for these efforts because modern systemic therapy, including endocrine treatment and CDK4/6 inhibitors for higher-risk disease, exerts systemic control regardless of nodal status, while the prognostic value of detecting a single micrometastasis has become increasingly debatable. A reliable preoperative risk estimate feeds directly into these shared decision-making conversations between surgeon and patient.

The study’s limitations are the ones that typically attend retrospective, single-institution models, and the authors are candid about them. All 540 patients came from one Madrid hospital, raising the possibility that referral patterns, imaging protocols, or pathology practices specific to that center could limit generalizability. The retrospective design means some variables, such as tumor-infiltrating lymphocytes or molecular assays not routinely ordered preoperatively, could not be incorporated. Most importantly, the model has not yet been tested on patients outside its development cohort. External validation, ideally in prospective multicenter cohorts drawn from different health systems, is required before any clinical implementation, and the authors state this requirement explicitly in their conclusions. History is littered with prediction models that performed beautifully in derivation and faded on external testing, which is why the field increasingly demands validation before publication of clinical tools is celebrated.

Even with those caveats, the contribution is timely and practical. Roughly one in three clinically node-negative HR+/HER2- patients in this cohort had occult nodal disease, and the strongest signal, the axillary ultrasound, is already obtained in most modern breast units, meaning the nomogram requires no new tests, no additional cost, and no delay to surgery. If external validation confirms the Madrid results, the tool could help identify the substantial minority of patients whose probability of nodal involvement is so low that even a sentinel procedure might reasonably be omitted, sparing them the attendant risks of lymphedema and nerve injury. Conversely, it could flag patients whose risk justifies a more thorough preoperative discussion about axillary management. In an era when precision oncology is often equated with genomic sequencing, this study is a reminder that careful statistical modeling of everyday clinical variables can still move the needle on surgical care, one calculated probability at a time.

Subject of Research: Development of a preoperative nomogram predicting axillary lymph node involvement in clinically node-negative HR+/HER2- breast cancer

Article Title: A preoperative nomogram to predict axillary lymph node involvement in clinically node-negative HR+/HER2- breast cancer

Article References: Carrillo-Peña, J., Osorio-Silla, I., Mahillo-Fernández, I., Sánchez De Molina-Rampérez, M. L., Escanciano-Escanciano, M., Pastor-Peinado, P., Guadalajara, H., García-Olmo, D., & Villarejo-Campos, P. (2026). A preoperative nomogram to predict axillary lymph node involvement in clinically node-negative HR+/HER2- breast cancer. BMC Cancer. https://doi.org/10.1186/s12885-026-17128-2

Image Credits: AI Generated

DOI: 10.1186/s12885-026-17128-2

Keywords: breast cancer, axillary lymph nodes, nomogram, sentinel lymph node biopsy, HR+/HER2-, surgical de-escalation, lymphovascular invasion, axillary ultrasound, prediction model, LASSO regression, risk stratification, BMC Cancer

A new study published in BMC Cancer by Jian Shang, Yaolu Li, Lidan Yang, Donghui Zhang and colleagues at the Affiliated Cancer Hospital of Guangzhou Medical University set out to change that timeline. The research team developed and internally validated a machine-learning model that combines standard clinical variables with quantitative features extracted from multiparametric magnetic resonance imaging to predict LVI before a patient ever enters the operating room. The work, a retrospective single-center study, also explored whether the model’s output could stratify patients by their risk of disease-free survival, offering a glimpse of how preoperative imaging intelligence might one day guide gastric cancer management.

The technical foundation of the study is radiomics, a field that converts the pixel-and-voxel patterns inside medical images into hundreds or thousands of numerical descriptors. Where a radiologist’s eye might notice that a tumor looks heterogeneous or ill-defined, a radiomics pipeline quantifies exactly how heterogeneous, how textured, how intense, and how variable the tumor’s signal is across different imaging sequences. The researchers drew on three complementary MRI sequences: fat-suppressed T2-weighted imaging, which highlights tissue water content and edema; apparent diffusion coefficient maps derived from diffusion-weighted imaging, which reflect the restricted movement of water molecules in densely packed tumor tissue; and venous-phase contrast-enhanced MRI, which reveals how the tumor takes up gadolinium-based contrast agent through its blood supply.

Each sequence captures a different facet of tumor biology. Restricted diffusion, for example, often corresponds to high cellularity, a hallmark of aggressive malignancies. Contrast enhancement patterns can reflect the chaotic, leaky neovasculature that tumors build to feed themselves. T2 signal characteristics can distinguish mucin-rich or necrotic regions from solid cellular components. By mining all three sequences simultaneously, the team aimed to build a digital fingerprint of the tumor that correlates with the microscopic reality of vascular invasion.

The study population comprised 458 patients with gastric cancer, randomly assigned to a training cohort of 320 patients and an internal test cohort of 138. This split is a critical methodological safeguard: models built and evaluated on the same patients routinely overstate their own accuracy, so holding out a test set provides a more honest estimate of performance. From the MRI data, the researchers extracted radiomics features and applied feature reduction techniques, including least absolute shrinkage and selection operator regression, known as LASSO, which penalizes complexity and drives the coefficients of uninformative features to zero. After this filtering process, six radiomics features and four clinical predictors survived as the final model inputs.

The team then constructed three logistic-regression models to compare different predictor sets, designating the combined clinical-radiomics model as the primary model. Logistic regression may seem modest in an era of deep neural networks, but it offers transparency, calibration, and robustness with modest sample sizes, qualities that matter enormously in clinical prediction. Secondary comparisons using alternative machine-learning algorithms, including support vector machines and linear discriminant analysis, were performed with fold-restricted random oversampling, a technique that balances the classes within each cross-validation fold to avoid data leakage while addressing the inevitable imbalance between LVI-positive and LVI-negative patients.

The headline result: the combined clinical-radiomics model achieved an area under the receiver operating characteristic curve, or AUC, of 0.775 in the training cohort and 0.752 in the internal test cohort, with a bootstrap 95 percent confidence interval of 0.663 to 0.840. An AUC of 0.75 places the model in the territory of moderate discrimination, meaning it performs meaningfully better than chance but is far from infallible. At a sensitivity-oriented threshold of 0.217, chosen in the training set using the Youden index, the model detected LVI with a test sensitivity of 0.795, catching roughly four out of five invasion-positive cases, while its negative predictive value reached 0.855. In practical terms, when the model says a tumor is unlikely to have lymphovascular invasion, that reassurance is correct about 86 percent of the time.

Calibration, the question of whether predicted probabilities match observed reality, is often neglected in machine-learning studies but received careful attention here. In the test cohort, the calibration intercept was 0.092, the slope 0.965, and the Brier score 0.180, all indicating that the model’s probability estimates were reasonably honest rather than systematically overconfident or underconfident. Decision-curve analysis further assessed the clinical utility of the model across a range of threshold probabilities, evaluating the net benefit a patient population would gain if decisions were made according to the model’s predictions.

Perhaps the most intriguing finding is exploratory: the model’s output appeared to carry prognostic weight. Using an X-tile-derived cutoff of 0.59 established in the training data and applied unchanged to the test cohort, the researchers divided patients into risk groups for disease-free survival. In an adjusted complete-case Cox proportional hazards model, each standard deviation increase in the model’s predicted probability was associated with a hazard ratio of 1.57 for worse disease-free survival, with a 95 percent confidence interval of 1.21 to 2.03. A single preoperative number, computed from scans and blood markers, appeared to encode information about how the disease would unfold, independent of the adjustment variables included in the analysis.

The authors are appropriately measured about what their model can and cannot do. They emphasize that the model is not ready for stand-alone clinical use and requires prospective external validation in independent centers before any deployment. The study was retrospective and single-center, meaning the MRI scanners, imaging protocols, and patient demographics of one institution shaped the data. Radiomics features are notoriously sensitive to variations in acquisition parameters, and a model trained in Guangzhou may not transfer cleanly to a hospital with different equipment. The disease-free survival findings, too, are explicitly labeled exploratory, and the sensitivity analyses using a training-median cutoff underscore the uncertainty around how the risk stratification should be operationalized.

Still, the study adds to a growing body of evidence that the information needed to characterize a tumor’s aggressiveness may already be hiding inside routine imaging. Every gastric cancer patient undergoing staging typically receives cross-sectional imaging; the marginal cost of running a radiomics algorithm on those existing scans is nearly zero. If future prospective studies confirm these results, a high negative predictive value at a sensitivity-oriented threshold suggests a plausible triage role: patients whose scans argue strongly against LVI might be candidates for less aggressive perioperative planning, while those flagged as high risk could be prioritized for intensified neoadjuvant therapy, extended lymph node dissection, or closer surveillance. The vision is not to replace pathologists but to move critical biological knowledge upstream in the treatment timeline, from the postoperative pathology report to the preoperative clinic visit.

The work also highlights a broader trend in oncology: the convergence of interpretable machine learning, multiparametric imaging, and clinical data into decision-support tools that are auditable rather than opaque. By retaining logistic regression as the primary model and reporting calibration alongside discrimination, the Guangzhou team modeled the kind of methodological discipline that regulatory science increasingly demands. The path from a 0.75 AUC in a single center to a validated clinical tool is long, but this study maps an early stretch of it, showing that the vessels a tumor invades in secret may leave traces visible to an algorithm long before any scalpel is lifted.

Subject of Research: Preoperative prediction of lymphovascular invasion in gastric cancer using multiparametric MRI radiomics and machine learning

Article Title: Preoperative prediction of lymphovascular invasion and prognostic risk stratification in gastric cancer using multiparametric MRI radiomics and machine learning: a retrospective model development and internal validation study

Article References: Preoperative prediction of lymphovascular invasion and prognostic risk stratification in gastric cancer using multiparametric MRI radiomics and machine learning: a retrospective model development and internal validation study. (n.d.). https://doi.org/10.1186/s12885-026-17042-7

Image Credits: AI Generated

DOI: 10.1186/s12885-026-17042-7

Keywords: gastric cancer, lymphovascular invasion, multiparametric MRI, radiomics, machine learning, logistic regression, disease-free survival, predictive modeling, medical imaging, cancer prognosis, internal validation, LASSO

A new retrospective study from Tehran University of Medical Sciences, published in BMC Cancer, now offers one of the largest and longest-window assessments of margin outcomes in a developing country to date. Drawing on more than two decades of clinical records, the research team led by Sadaf Alipour, Bita Eslami, Nasim Nouri, Ramesh Omranipour, and Samareh Heydari analyzed the cases of 2,585 women who underwent breast-conserving surgery between 2003 and 2024. The scale of the dataset matters: with a median patient age of 49 years and a mean tumor diameter of 27 millimeters, the cohort captures the typical profile of breast cancer presentation in Iran, where disease often appears in younger women and at somewhat larger sizes than in Western registries.

The headline finding is unexpectedly encouraging. Histological examination of the surgical specimens revealed free, or negative, margins in 91.4 percent of cases, meaning that only 8.6 percent of patients had cancer cells extending to the edge of the removed tissue. That figure is lower than the rates reported in much of the existing international literature, where margin positivity after breast-conserving surgery has commonly been described at rates ranging from roughly 10 to 30 percent depending on the population, the definition of an adequate margin, and the era of the study. For a resource-constrained setting often assumed to face structural disadvantages in surgical oncology, the result challenges a persistent assumption that outcomes in developing countries must lag behind those of wealthier health systems.

The technical stakes of margin status are worth spelling out. When a pathologist examines the excised lumpectomy specimen, they ink the outer surfaces and section the tissue to determine whether tumor cells touch the inked edge. A positive margin means residual disease may remain in the breast, and the standard response is re-excision—a second operation to remove additional tissue from the involved cavity wall. Beyond the immediate burden of repeat anesthesia, hospitalization, and scarring, positive margins are associated with elevated rates of local recurrence, since any remaining tumor cells can seed regrowth. Guidelines from major surgical societies therefore treat a negative margin as a non-negotiable endpoint of breast-conserving therapy, and quality metrics in many countries track institutional re-excision rates as a proxy for surgical excellence.

Every patient in the Iranian cohort who had an involved margin underwent re-excision of the affected edge, the authors report, indicating that the clinical pathway for managing positive margins was consistently followed. That consistency itself is informative. In some low- and middle-income settings, fragmented referral pathways or limited pathology capacity can delay or complicate the identification of margin involvement. The Tehran experience suggests that a large academic center with integrated surgical and pathological services can achieve both a low primary positivity rate and reliable correction when positivity occurs, providing a template for other institutions in comparable health systems.

Perhaps the most clinically useful contribution of the study is its identification of preoperative predictors of margin involvement. Using both univariate and multivariate logistic regression models, the researchers examined how a range of tumor and patient characteristics related to the odds of a positive margin. Three factors emerged as significantly associated: younger patient age, larger tumor size, and the presence of lymphovascular invasion, the histological finding of tumor cells within small blood or lymphatic vessels surrounding the tumor. Each of these variables can be assessed before or at the time of surgery, which is precisely what makes them actionable.

The biology behind these associations is plausible. Younger patients tend to have denser breast parenchyma and, in some populations, more biologically aggressive tumor subtypes, which can make it harder for surgeons to gauge the extent of disease by palpation or imaging. Larger tumors occupy more volume relative to the breast and are more likely to have irregular, infiltrative borders that extend beyond what preoperative measurements suggest. Lymphovascular invasion, meanwhile, is a recognized marker of invasive behavior, indicating a propensity for tumor cells to spread along tissue planes—exactly the pattern that produces microscopic extensions past a grossly clear resection line. A surgeon who knows, before making the first incision, that a patient carries these risk factors can plan wider excision margins, request intraoperative assessment, or counsel the patient about the possibility of a second procedure.

The study’s methodology deserves attention for what it says about the reliability of the findings. By restricting inclusion to female patients with complete data and spanning a 21-year period, the investigators minimized the selection biases that often plague single-center surgical series. The long time window also implicitly captures the evolution of breast surgical technique over two decades, including the gradual adoption of oncoplastic approaches, improved preoperative imaging, and more standardized pathology protocols. The research was approved by the ethics committee of Tehran University of Medical Sciences and conducted with written informed consent under the Declaration of Helsinki, and the authors declare no competing interests, with funding provided by the university’s Deputy of Research.

For the global surgical oncology community, the implications cut in two directions. First, the data provide a benchmark against which other centers in developing countries can measure their own performance, filling a documented evidence gap that has made it difficult to design regionally appropriate quality-improvement programs. Second, the identified predictors give clinicians in any setting a simple, low-cost risk stratification tool: age, tumor size, and lymphovascular status are all routinely available from standard workups, requiring no advanced technology to apply. In health systems where re-operation carries disproportionate financial and logistical burdens for patients—many of whom travel long distances for care—preoperative risk prediction has an especially high practical value.

The study also serves as a reminder that surgical quality is not a function of national income alone. With careful technique, integrated pathology, and attention to known risk factors, breast-conserving surgery in a developing country can achieve margin outcomes that compare favorably with the best published figures. As the authors conclude, recognizing the clinical and pathological factors linked to margin involvement allows for better risk prediction and potential surgical adjustment before the operation begins—turning what has historically been a reactive, post-hoc problem into one that can be anticipated and, increasingly, prevented.

Subject of Research: Positive surgical margin rates and predictors in breast-conserving surgery for breast cancer in a developing country

Article Title: Frequency of positive margins and predictors in breast-conserving surgery in developing countries

Article References: Frequency of positive margins and predictors in breast-conserving surgery in developing countries. (n.d.). https://doi.org/10.1186/s12885-026-17073-0

Image Credits: AI Generated

DOI: 10.1186/s12885-026-17073-0

Keywords: breast cancer, breast-conserving surgery, positive surgical margin, re-excision, lymphovascular invasion, surgical oncology, Iran, risk factors, local recurrence, developing countries, tumor size, pathology

The review, led by Yassin Rahnama and colleagues at Tehran University of Medical Sciences, set out to quantify just how well AI models—spanning both handcrafted radiomics and end-to-end deep learning—can predict LVI from preoperative medical imaging such as magnetic resonance imaging, computed tomography, and positron emission tomography. The team searched four major biomedical databases, PubMed, Embase, Scopus, and Web of Science, with coverage extending to October 3, 2025. To be included, studies had to involve patients with histopathologically confirmed colorectal cancer who underwent preoperative imaging that was subsequently analyzed by AI algorithms, with the final diagnosis of LVI established by postoperative pathology serving as the reference standard.

After screening, the researchers assembled 18 retrospective studies encompassing a combined 5,524 patients. Seventeen of those studies reported enough data to be folded into a quantitative meta-analysis. The pooled results were encouraging: sensitivity came in at 0.77, with a 95 percent confidence interval of 0.70 to 0.82, meaning the models correctly flagged roughly three out of four tumors that truly harbored LVI. Specificity was slightly higher at 0.81, with a confidence interval of 0.77 to 0.85, indicating that the algorithms also did a credible job of ruling out invasion when it was absent. The summary receiver operating characteristic curve, a standard way of visualizing diagnostic performance across studies, yielded an area under the curve of 0.84, with a partial AUC of 0.74—a figure the authors characterize as robust for a diagnostic task of this difficulty.

Those headline numbers become more tangible when translated into clinical terms. At a pre-test probability of 38 percent, a plausible prevalence of LVI in a typical colorectal cancer population, the pooled estimates correspond to a positive predictive value of 72 percent and a negative predictive value of 85 percent. In practical terms, a positive AI prediction would meaningfully raise suspicion that a tumor has invaded lymphatic or vascular channels, while a negative prediction would be reassuring in the large majority of cases. That kind of preoperative risk stratification could influence surgical planning, the extent of lymph node dissection, and the timing of conversations about adjuvant therapy—all before the patient is ever anesthetized.

The technical machinery behind these predictions is worth unpacking. Radiomics approaches convert medical images into hundreds or thousands of quantitative features describing tumor shape, texture, intensity, and heterogeneity, which are then fed into machine learning classifiers such as support vector machines or random forests. Deep learning approaches, by contrast, use convolutional neural networks that learn discriminative patterns directly from the pixel data, often capturing subtle image signatures that escape both human perception and manually engineered features. The fact that both families of algorithms converge on similar pooled accuracy suggests that the imaging phenotype of LVI—presumably reflecting tumor biology such as desmoplastic reaction, peritumoral fat infiltration, and altered vascular patterns—carries genuine, extractable signal.

Quality assessment, however, tempers the enthusiasm. The authors applied the QUADAS-2 tool, the standard framework for evaluating bias and applicability in diagnostic accuracy studies, alongside the Radiomics Quality Score 2.0, or RQS 2.0, which grades how rigorously radiomics research is conducted. The mean RQS 2.0 score across the included studies was just 24.7 out of a possible 100 percent scale—equivalent to 44 percent—highlighting substantial methodological shortcomings in the existing literature. Common weaknesses in this field include small single-center cohorts, retrospective designs vulnerable to selection bias, inconsistent image acquisition protocols, and limited reporting of model validation. These deficiencies mean the impressive pooled statistics should be read as a proof of concept rather than a guarantee of real-world performance.

Geography presents another significant caveat. Every one of the 18 included studies was retrospective, and all were conducted in China. That homogeneity raises genuine questions about generalizability. Imaging protocols, scanner vendors, patient demographics, tumor biology, and even the prevalence of LVI itself can differ across populations and healthcare systems, and an algorithm trained on one distribution of data may degrade when deployed in another. The authors are explicit that prospective, multicenter validation in geographically and ethnically diverse populations, together with standardized imaging protocols, is required before broader clinical implementation can be considered. Without such validation, the pooled estimates remain a portrait of what is achievable under research conditions, not what patients should expect at their local hospital.

Even so, the clinical logic of the endeavor is compelling. Current practice leaves surgeons operating on colorectal cancer essentially blind to LVI status, relying on intraoperative judgment and postoperative pathology to guide treatment intensity. A validated AI tool that reads routine MRI or CT scans—imaging most patients already receive during staging—could slot into existing workflows without additional radiation, cost, or invasive procedures. It could help identify patients who might benefit from more extensive lymphadenectomy, flag candidates for neoadjuvant therapy, or enrich clinical trials by preselecting patients at higher risk of occult spread. The negative predictive value of 85 percent, in particular, hints at a role in reassuring both clinicians and patients when the algorithm finds no evidence of vascular invasion.

The study also fits into a broader movement in oncology imaging, where machine learning is being trained to predict molecular and histological features—microsatellite instability, tumor-infiltrating lymphocytes, genetic mutations—directly from radiographic appearances, a paradigm sometimes called radiogenomics. LVI prediction is a natural extension of that agenda because it targets a well-validated prognostic marker with a clear pathological definition and direct treatment implications. The convergence of computer vision, growing archives of digitized pathology-confirmed scans, and standardized reporting frameworks like QUADAS-2 and RQS 2.0 is steadily raising the bar for what counts as credible evidence in this space.

The path from meta-analysis to clinic remains long, but the trajectory is clear. The Tehran-led team, whose work was conducted under PRISMA reporting guidelines and published open access, has provided the most comprehensive quantitative synthesis to date of AI-driven LVI prediction in colorectal cancer imaging. Their verdict is measured: pooled sensitivity of 0.77, specificity of 0.81, and an AUC of 0.84 constitute promising diagnostic accuracy that may serve as a complementary tool for preoperative risk stratification. What happens next depends on whether the field can graduate from retrospective, single-country studies to the kind of prospective, externally validated, protocol-standardized trials that turn clever algorithms into trusted clinical instruments. For the thousands of patients diagnosed with colorectal cancer each year, the prospect of knowing, before surgery, whether their tumor has already begun its silent march through the vasculature is a prize well worth the rigorous validation still to come.

Subject of Research: Artificial intelligence for preoperative prediction of lymphovascular invasion in colorectal cancer imaging

Article Title: Artificial intelligence in predicting lymphovascular invasion in colorectal cancer imaging: a systematic review and meta-analysis

Article References: Rahnama, Y., Ahmadi-Tafti, S. M., Yousefi-Koma, H., Pirahesh, K., Torkaman, N., BaradaranRad, A., Keshvari, A., Salahshour, F., Sedaghat, M., & Delazar, S. (2026). Artificial intelligence in predicting lymphovascular invasion in colorectal cancer imaging: a systematic review and meta-analysis. BMC Medical Imaging. https://doi.org/10.1186/s12880-026-02858-3

Image Credits: AI Generated

DOI: 10.1186/s12880-026-02858-3

Keywords: colorectal cancer, lymphovascular invasion, artificial intelligence, radiomics, deep learning, meta-analysis, medical imaging, MRI, diagnostic accuracy, risk stratification, predictive markers, BMC Medical Imaging

Triple-negative breast cancer is defined by the absence of the three molecular handles that anchor most modern breast cancer therapy: the estrogen receptor, the progesterone receptor, and HER2. Without those targets, endocrine agents and HER2-directed drugs are ineffective, leaving cytotoxic chemotherapy as the systemic backbone of treatment. Although triple-negative tumors account for only about 10 to 15 percent of breast cancers, they carry a disproportionate share of the disease’s mortality because they tend to grow quickly, are more likely to metastasize, and historically have fewer therapeutic options. For patients whose tumors are caught early, the central treatment question is not whether chemotherapy helps but when it should be given.

That question is most acute for T1c tumors, which measure between one and two centimeters and have not spread to regional lymph nodes or distant sites. The T1cN0M0 designation describes a small, node-negative cancer, yet even at this stage triple-negative biology demands systemic therapy to reduce recurrence risk. The debate concerns sequencing. Neoadjuvant chemotherapy, delivered before surgery, offers the chance to shrink the tumor, potentially enabling less extensive operations, and provides real-time information about how the cancer responds to treatment. Adjuvant chemotherapy, given after surgery, achieves the same cytotoxic goal but leaves the surgeon operating on a tumor that has never been exposed to drugs. National guidelines leave room for either approach in this small-tumor setting, and practice varies considerably across institutions.

The reason sequencing remains contested is that neither strategy has demonstrated superior overall survival. Multiple prior analyses, including studies of the same National Cancer Database, have found that patients with clinically node-negative T1 triple-negative tumors fare equally well whether chemotherapy comes first or last. If survival outcomes are equivalent, the choice between the two approaches hinges on secondary benefits: the possibility of breast-conserving surgery, the prognostic information embedded in treatment response, and patient preference. The Johns Hopkins team, led by breast surgical oncology researcher Andrew Venardi, set out to sharpen that calculus by asking a more granular question: within this seemingly uniform early-stage population, which patients are actually likely to benefit from going first with chemotherapy?

The answer, they hypothesized, might lie in Ki-67, a nuclear protein expressed by cells that are actively dividing. Pathologists measure Ki-67 by immunohistochemistry, staining tumor samples and counting the percentage of positive cells, which yields a proliferation index reflecting how fast the cancer is growing. The logic connecting Ki-67 to neoadjuvant strategy is mechanistically straightforward: chemotherapy agents such as anthracyclines and taxanes preferentially kill rapidly proliferating cells, so tumors with a high fraction of dividing cells should, in principle, be more sensitive to upfront drug treatment. Previous studies have associated high Ki-67 expression with better responses to neoadjuvant chemotherapy in triple-negative disease, but the specific threshold that separates responders from non-responders in the earliest tumor category had not been clearly defined using large-scale national data.

To conduct the analysis, the researchers queried the National Cancer Database for female patients aged 18 and older diagnosed with T1N0M0 triple-negative breast cancer between 2018 and 2022 who underwent both chemotherapy and surgery. The National Cancer Database, jointly maintained by the American College of Surgeons and the American Cancer Society, captures roughly 70 percent of newly diagnosed cancer cases in the United States, providing the statistical power that single-institution series cannot match. Patients were divided into a neoadjuvant chemotherapy cohort and an adjuvant chemotherapy cohort. Because the two groups differ systematically in tumor size, grade, and other characteristics—surgeons and oncologists tend to select patients for upfront chemotherapy based on perceived risk—the team applied inverse probability of treatment weighting, a statistical technique that re-weights the cohorts to mimic the balance of a randomized trial.

The survival analysis produced a finding consistent with the existing literature: overall survival did not differ significantly between the neoadjuvant and adjuvant groups. Multivariate Cox regression, which adjusts for competing variables simultaneously, instead identified three factors associated with worse survival. Lymphovascular invasion, the presence of tumor cells within small blood or lymphatic vessels, signaled a higher risk of microscopic spread. The invasive lobular subtype, which arises in the milk-producing lobules rather than the ducts, and mixed lobular-ductal tumors were also linked to poorer outcomes. Lobular triple-negative cancers are rare and biologically distinct, and recent registry-based studies have shown they can carry worse survival despite appearing indolent, so their emergence as an adverse marker in this dataset aligns with a growing body of evidence.

The most consequential result came from the response analysis. Among patients who received neoadjuvant chemotherapy, those who achieved a pathologic complete response—meaning no residual invasive cancer was found in the breast or lymph nodes at the time of surgery—had markedly better survival than those who did not. Non-responders faced a hazard ratio of 3.59 for worse overall survival compared with patients who achieved a complete response. This echoes the findings of a large meta-analysis showing that pathologic complete response in triple-negative breast cancer is strongly associated with improved long-term event-free and overall survival. In other words, while the timing of chemotherapy may not change survival on average, achieving eradication of the tumor before surgery identifies a subgroup whose prognosis is transformed.

That is where the Ki-67 threshold enters. Using a receiver operating characteristic curve, the researchers plotted the relationship between Ki-67 expression and pathologic complete response and applied the maximum Youden index, a standard method for finding the cutoff point that best balances sensitivity and specificity. The threshold that emerged was 45.8 percent. Patients whose tumors expressed Ki-67 at or above that level achieved a pathologic complete response rate of 45.3 percent, while those below it responded at a rate of only 21.8 percent. Expressed differently, a high proliferation index more than doubled the odds that upfront chemotherapy would completely eliminate the tumor. Logistic regression confirmed Ki-67 expression as an independent predictor of complete response, suggesting the association held after accounting for other tumor characteristics.

The clinical implications are tangible for a patient population that often faces an anxious choice. A woman with a 1.5-centimeter, node-negative triple-negative tumor whose biopsy shows a Ki-67 index of 50 percent now has registry-scale evidence that chemotherapy before surgery is likely to produce a complete response, with the dual benefits of maximal prognostic information and a survival outcome that appears markedly better than that of non-responders. Conversely, a patient with a low proliferation index may reasonably opt for surgery first, since her probability of achieving a complete response is lower and adjuvant chemotherapy delivers equivalent survival. The authors emphasize that the study is retrospective and that Ki-67 measurement itself has well-known variability between laboratories and staining protocols, so the 45.8 percent cutoff should be validated prospectively before it reshapes guidelines. Still, as the field moves from one-size-fits-all recommendations toward personalized sequencing in stage I triple-negative breast cancer, a routine pathology marker already on every diagnostic report may prove to be the deciding variable.

Subject of Research: Ki-67 expression as a predictor of pathologic complete response to neoadjuvant chemotherapy in early-stage triple-negative breast cancer

Article Title: Ki-67 expression and rates of pathologic complete response in T1cN0M0 triple-negative breast cancer: a national cancer database analysis

Article References: Venardi, A., Shojaeian, F., Diaz, S., Schuster, C. R., Rath, P., Shaid, I., Singh, A., Santa-Maria, C., & Sogunro, O. (2026). Ki-67 expression and rates of pathologic complete response in T1cN0M0 triple-negative breast cancer: a national cancer database analysis. Breast Cancer Research and Treatment, 219(3), Article 21. https://doi.org/10.1007/s10549-026-08086-1

Image Credits: AI Generated

DOI: 10.1007/s10549-026-08086-1

Keywords: triple-negative breast cancer, Ki-67, neoadjuvant chemotherapy, pathologic complete response, National Cancer Database, T1cN0M0, proliferation index, biomarker, breast surgical oncology, lymphovascular invasion, invasive lobular carcinoma, personalized treatment

LVI describes the presence of tumor cell clusters inside endothelial-lined lymphatic vessels or blood vessels surrounding an invasive breast cancer. It is an independent predictor of local recurrence, distant metastasis, and poor prognosis, yet confirming it today depends entirely on postoperative pathology. Surgeons learn whether tumor cells had already entered vessels only after the tumor has been removed, which limits their ability to fine-tune the extent of axillary lymph node dissection or to personalize adjuvant treatment decisions. A reliable preoperative, noninvasive estimate of LVI status has therefore been a long-standing unmet need in breast oncology.

The new study tackles that gap using dynamic contrast-enhanced magnetic resonance imaging, or DCE-MRI, already a core preoperative modality thanks to its exceptional soft-tissue resolution and its sensitivity to tumor vascularization. The researchers combined DCE-MRI with radiomics, a technique that extracts large numbers of quantitative image features invisible to the human eye, and with machine learning, allowing algorithms to decode tumor heterogeneity from routine scans. While earlier efforts have used mammography-based tomosynthesis or single-center MRI cohorts, systematic comparisons of many algorithms on multicenter DCE-MRI data, with rigorous external validation, remained scarce.

Retrospectively, the team assembled 912 female patients aged 23 to 81 years from two independent medical centers, Guangdong Maternal and Child Health Hospital and The First Affiliated Hospital of Jinan University, all scanned with a Philips 3.0T scanner within two weeks before surgery. Data from 757 patients at the first center formed the model development set, split 7:3 into training and internal validation subsets using stratified sampling on LVI status. Data from 155 patients at the second center served as a completely unseen external test set, a design that directly probes how well the model generalizes to a different hospital population rather than merely memorizing patterns from its own data.

The radiomics pipeline began with manual, blinded contouring of the entire tumor on the arterial phase of the six-phase contrast sequence, producing a three-dimensional volume of interest reviewed by senior radiologists. Using PyRadiomics, the team extracted 1,197 features per patient, spanning 234 histogram, 14 morphological, 884 texture, and 65 higher-order descriptors. A multistep reduction process followed: correlation filtering removed redundant features, univariate analyses retained only those significantly associated with LVI, and LASSO regression with 10-fold cross-validation ultimately selected 18 features with the highest predictive value for modeling.

Crucially, the researchers did not commit to a single algorithm. They built and systematically compared 10 machine learning models, including logistic regression, support vector machine, K-nearest neighbors, random forest, ExtraTrees, XGBoost, LightGBM, gradient boosting, AdaBoost, and multilayer perceptron, tuning each via grid search and comparing them with ROC analysis and the DeLong test. The ExtraTrees model emerged as the best radiomics performer, achieving an AUC of 0.653 and accuracy of 0.640 on the validation set, a deliberately conservative benchmark compared with many single-center studies that report higher numbers without independent external testing.

In parallel, multivariate logistic regression identified three independent clinical risk factors for LVI: sentinel lymph node status, estrogen receptor status, and the time-intensity curve pattern of contrast enhancement. The team then integrated the ExtraTrees radiomics score with these clinical predictors to construct a combined model, visualized as a practical nomogram for individualized bedside estimation of LVI probability. The combined model outperformed either the radiomics or clinical model alone across all cohorts, reaching an AUC of 0.796 on the training set, 0.704 on internal validation, and 0.703 on the external test set, with calibration curves and decision curve analysis confirming good agreement and clinical net benefit across a broad range of threshold probabilities.

The authors argue that their relatively modest external AUC values are not a weakness but a feature of honest validation. Many published radiomics studies reporting AUCs above 0.80 relied on single-center data, lenient validation schemes, or highly homogeneous patient subsets. By enrolling all molecular subtypes, including luminal A, luminal B, HER2-overexpressing, and triple-negative cancers, and by testing on a genuinely independent center with real-world data heterogeneity, the study offers a more realistic assessment of how such models behave outside their birthplace. The researchers also note that no dedicated image harmonization strategy, such as ComBat, was applied before feature extraction, which may partly explain the performance drop on external data and marks a target for future improvement.

Limitations remain. The retrospective design carries inherent selection bias, the external cohort was relatively small, and tumor delineation still depends on manual contouring by experienced radiologists, a process that is time-consuming and subject to inter-observer variability. The two centers also used the same scanner vendor, and prospective validation in larger, more diverse populations will be essential before routine clinical deployment. The team outlines next steps that include incorporating diffusion-weighted MRI parameters, applying explainable AI to reveal the imaging basis of model decisions, and integrating the tool into clinical decision support systems to measure its actual impact on surgical planning and patient outcomes.

Even with those caveats, the study delivers a clinically meaningful proof of concept: a noninvasive, preoperative estimate of lymphovascular invasion is achievable when carefully selected radiomic features from DCE-MRI are combined with independent clinical risk factors and validated the hard way. For patients, that could mean smarter surgical decisions and more personalized adjuvant therapy planned before the operation rather than pieced together after it. For the growing field of radiomics, it is a reminder that multicenter external validation, systematic algorithm comparison, and honest reporting of performance are what separate a promising research model from one ready for the clinic.

Subject of Research: Preoperative prediction of lymphovascular invasion in breast cancer using DCE-MRI radiomics and machine learning

Article Title: Machine Learning Comparison and Combined Model Optimization Based on DCE‐MRI Radiomics for Preoperative Assessment of Lymphovascular Invasion in Breast Cancer: A Multicenter Study

Article References: Li, H., Xiao, Q., Zeng, Y., Wang, X., & Zhang, Y. (2026). Machine Learning Comparison and Combined Model Optimization Based on DCE ‐ MRI Radiomics for Preoperative Assessment of Lymphovascular Invasion in Breast Cancer: A Multicenter Study. Cancer Reports, 9(9), Article e70678. https://doi.org/10.1002/cnr2.70678

Image Credits: AI Generated

DOI: 10.1002/cnr2.70678

Keywords: breast cancer, lymphovascular invasion, DCE-MRI, radiomics, machine learning, ExtraTrees, nomogram, multicenter study, external validation, clinical risk factors, LASSO regression, preoperative assessment

The research, conducted at Hospital Canselor Tuanku Muhriz of Universiti Kebangsaan Malaysia and published in BMC Cancer, examined 241 women with clinically node-negative early breast cancer, all staged cT1–2N0M0, who underwent sentinel lymph node biopsy between January 2018 and December 2025. The team, led by surgeon Hadija Mustamin together with pathologists and community health researchers, asked a deceptively simple question: does the density of tumour-infiltrating lymphocytes, or TILs, in the tumour stroma correlate with whether cancer cells are found in the sentinel node?

The findings were striking. Among the cohort, 67 patients — 27.8 percent — had metastasis to the sentinel lymph node. But the distribution of that risk was far from even. When pathologists graded stromal TIL levels according to the internationally accepted 2014 guidelines of the International TILs Working Group, they found that two-thirds of the patients fell into the low-TIL category, 17.4 percent into the intermediate category, and 16.6 percent into the high category. Critically, the highest proportion of nodal metastasis occurred in the group with the fewest infiltrating lymphocytes, a statistically significant association with a p value of 0.023.

To understand why this matters, one must appreciate what TILs actually represent. Tumour-infiltrating lymphocytes are immune cells — predominantly T cells — that have migrated from the bloodstream into the tumour microenvironment, a biological indication that the host immune system has recognised the malignancy and mounted a response. In recent years, TIL assessment has graduated from a curiosity of tumour immunology to a practical, reproducible pathology metric. In triple-negative and HER2-positive breast cancers, high TIL counts are already associated with better responses to neoadjuvant chemotherapy and improved survival. What this new study adds is a potential role in the preoperative and perioperative decision-making around the axilla.

The statistical backbone of the study was a two-stage logistic regression strategy. The researchers first screened clinicopathological variables — tumour size, histological type, grade, lymphovascular invasion, hormone receptor status and TIL category — using univariable analyses, and then entered any variable with a p value below 0.25 into a multivariable model. After adjustment, two factors emerged as independent predictors of sentinel node involvement. Higher TIL levels were associated with lower odds of metastasis, with an adjusted odds ratio of 0.977 per unit increase (95 percent confidence interval 0.961–0.994; p = 0.007). Lymphovascular invasion, the histological evidence of tumour cells invading blood or lymphatic vessels, was independently associated with nodal spread, carrying an adjusted odds ratio of 0.290 (95 percent CI 0.156–0.537; p < 0.001), meaning its presence substantially raised the odds of metastasis.

Interestingly, tumour size — long regarded as a cornerstone of nodal risk prediction — was associated with sentinel lymph node metastasis in the univariable analysis but lost its significance once other factors were accounted for. That statistical humility is itself informative: it suggests that within this cohort of small, clinically node-negative tumours, the biology of the tumour microenvironment and the presence of lymphovascular invasion may carry more predictive weight than size alone. If validated, such a shift could change how pathologists’ reports are weighted in surgical planning.

The clinical implications are potentially significant. Sentinel lymph node biopsy, while far less morbid than full axillary dissection, still carries risks of lymphoedema, seroma, nerve injury and infection. A reliable, cheap, microscope-based biomarker that stratifies nodal risk before surgery could, in principle, help identify patients in whom biopsy yields the greatest information, or in the longer term, support more selective approaches to axillary staging. Because TIL scoring requires only routine haematoxylin and eosin slides and a trained pathologist, it is vastly more accessible than genomic assays or advanced imaging, an especially important consideration in resource-constrained health systems such as much of Southeast Asia.

The authors are careful, and appropriately so, about the limits of their work. This was a single-centre, cross-sectional study, which captures association but cannot establish that low TIL levels cause nodal spread. Only a single pathologist scored the slides, raising questions about inter-observer reproducibility that multi-reader studies would need to resolve. The cohort spans tumours of all molecular subtypes, and TIL biology differs markedly between triple-negative, HER2-positive and luminal cancers; subtype-specific analyses in larger, prospective cohorts with external validation will be required before TIL assessment can be folded into clinically useful prediction models for nodal status. The study received no external funding, and the analysis reflects institutional records from one Malaysian tertiary centre.

Nevertheless, the study joins a rapidly expanding body of evidence that the immune microenvironment is not merely a bystander in cancer progression but an active determinant of how tumours behave. A tumour bristling with lymphocytes appears, in this data, to be a tumour less likely to have seeded the regional nodes — a biological narrative that makes intuitive sense, since a robust local immune response may constrain the earliest steps of dissemination. Conversely, an immunologically cold tumour, walled off from immune attack, may enjoy freer rein to invade lymphatic channels and establish footholds downstream.

For patients, the message is not one of immediate change in care — no guideline today spares anyone a sentinel node biopsy on the basis of TIL counts. But it is another step toward a future in which the pathology report of a breast cancer describes not just what the tumour is, but what the body is doing about it. As immunotherapy reshapes oncology and immune biomarkers move from research benches to clinical reports, studies like this one map the terrain where tumour immunology and surgical decision-making converge. Whether the quiet presence of lymphocytes in a tumour’s stroma can one day spare a woman an unnecessary procedure remains to be proven, but this Malaysian cohort offers a compelling early signal that the immune system has been keeping records all along — records pathologists are only now learning to read.

Subject of Research: Association between stromal tumour-infiltrating lymphocyte levels and sentinel lymph node metastasis in early breast cancer

Article Title: Tumour-infiltrating lymphocytes are associated with sentinel lymph node metastasis in early breast cancer: a cross-sectional study

Article References: Tumour-infiltrating lymphocytes are associated with sentinel lymph node metastasis in early breast cancer: a cross-sectional study. (n.d.). https://doi.org/10.1186/s12885-026-17012-z

Image Credits: AI Generated

DOI: 10.1186/s12885-026-17012-z

Keywords: breast cancer, tumour-infiltrating lymphocytes, sentinel lymph node metastasis, tumour immune microenvironment, lymphovascular invasion, sentinel lymph node biopsy, cancer metastasis, tumour immunology, early breast cancer, pathology biomarker, logistic regression, BMC Cancer