Evaluating AI and Manual Stromal Immune Cell Quantification in HER2 Positive Breast CancerAI tools help predict outcomes for HER2-positive breast cancer

The Lancet. Oncology Published September 10, 2026 Study authors: Lara González Luis E, Giudici Fabiola, Salgado Roberto, Yan Haixi, Rojo Federico, Comerma Laura, Wie… PubMed ↗ DOI ↗ Editorial oversight: Dr. Julia Lee, PhD · Oncology, Genomics & Drug Development

AI-generated summary of the cited source, checked by automated accuracy review. How we work

Key Takeaway

Higher sTIL levels correlate with better outcomes, and pertuzumab shows significant benefit in patients with high sTIL.

This secondary analysis of a Phase 3 randomized trial investigates the utility of various stromal immune lymphocyte (sTIL) quantification methods in patients with early-stage HER2-positive breast cancer. The study evaluated manual scoring, digital imaging, and AI-based metrics to determine their reliability and clinical utility in predicting invasive disease-free survival (IDFS) and identifying patients who benefit from pertuzumab addition to trastuzumab.

Manual scoring demonstrated high interobserver reproducibility, evidenced by an ICC of 0.84. This suggests that while manual methods are standard, they provide a reliable baseline for clinical assessment. The study specifically looked at how these measurements correlate with patient outcomes over a median follow-up of approximately 74 months, providing a robust dataset for evaluating prognostic indicators.

Analysis of IDFS revealed that higher levels of TILs, regardless of the measurement method used, were associated with improved survival outcomes. Specifically, the data indicated that patients with higher sTIL counts experienced better clinical trajectories. This reinforces the role of the tumor microenvironment as a critical prognostic factor in HER2-positive breast cancer management.

When evaluating the impact of pertuzumab, the study found that the addition of pertuzumab was associated with improved IDFS specifically in patients with higher sTIL levels across all measurement modalities. This suggests that sTIL quantification can serve as a predictive biomarker to identify patients most likely to derive a significant clinical benefit from the pertuzumab-trastuzumab regimen.

In node-positive patients with high manual sTIL scores (greater than or equal to 70.0%), the addition of pertuzumab resulted in a 12.1 percentage point absolute improvement in 6-year survival. This finding highlights the importance of high-quality sTIL quantification in identifying high-risk patients who may benefit from intensified systemic therapy.

AI-based immune hotspot scores, when integrated into nested models, provided the most consistent additional information when combined with any sTIL measurement method. These spatial metrics offered complementary data that enhanced the predictive accuracy of the models. While the concordance between manual and automated methods was noted as modest, the AI-derived metrics provided unique insights into the spatial distribution of immune cells. Clinicians can utilize these findings to better stratify patients based on their immune profile. While manual counting remains a reliable standard, AI-based spatial metrics offer a sophisticated layer of data that may refine treatment decisions. Future studies are needed to validate these AI-driven metrics in independent cohorts to establish them as standard clinical tools.

How this fits prior evidence

This secondary analysis of the APHINITY trial extends prior coverage on risk stratification in breast cancer. It aligns with the AI-enabled D-dimeromics finding by demonstrating that AI-based spatial metrics can provide complementary prognostic information, though both require clinical validation. The high reproducibility of manual sTIL scoring (ICC 0.84) contrasts with the modest concordance between manual and automated methods, echoing the need for validation seen with shear wave elastography. Unlike the exercise and cardiac troponin study, this analysis focuses on immune markers rather than cardioprotection, but similarly highlights the potential of adjunctive biomarkers to refine treatment decisions.

For people living with early-stage HER2-positive breast cancer, knowing how the body's immune system reacts to the disease is a vital piece of the puzzle. One specific marker, called sTILs, refers to immune cells that infiltrate the tumor. Understanding these cells helps doctors gauge how well a patient might respond to specific treatments, such as the combination of trastuzumab and pertuzumab. This research looks at how different ways of counting these cells, including new computer-aided methods, can help predict a patient's journey.

Researchers looked at data from a large trial involving over 4,800 patients. They compared three different ways to measure these immune cells: manual counting by experts, digital counting, and advanced AI-based methods. They also looked at spatial metrics, which are ways to measure where these cells are located within the tumor. The goal was to see if these different measurement styles could accurately predict how long a patient would remain free from invasive disease.

The results showed that higher levels of these immune cells were linked to better outcomes for patients. Specifically, when patients with a high number of these cells were treated with pertuzumab, they showed a significant improvement in staying disease-free. In a specific group of patients with lymph node involvement and high manual counts of these cells, there was a 12.1 percentage point improvement in survival over six years. Additionally, the AI-based tools provided consistent information that helped clarify the outlook for patients when combined with other measurement methods.

While the results are promising, there are important things to keep in mind. The study noted that while the AI tools were consistent, there was only a modest agreement between the manual and automated counting methods. This means that while the technology is helpful, it is not yet a perfect replacement for human experts. Also, the study was a secondary analysis of a larger trial, which means it is one piece of the puzzle rather than a brand-new clinical trial.

For patients today, this means that while these AI tools are not yet the standard of care, they show great potential for the future. They could eventually help doctors more accurately predict which patients will benefit the most from specific drug combinations. For now, these findings provide a clearer picture of how immune cell counts can serve as a helpful guide in personalizing treatment plans for breast cancer.

What this means for you:

AI and manual counts of immune cells can help predict how well some breast cancer patients respond to treatment.

Study Details

Study typeRct

Sample sizen = 4,805

EvidenceLevel 2

Follow-up1.0 mo

PublishedSep 2026

PMID42636838

View Original Abstract ↓

BACKGROUND: Stromal tumour-infiltrating lymphocytes (sTILs) are prognostic in early-stage HER2-positive breast cancer, but their role in the context of dual HER2 blockade remains undefined. We evaluated manual, digital, and artificial intelligence (AI)-based sTIL quantification, together with AI-derived spatial metrics, for prognostic and treatment-benefit stratification using tumour samples from the phase 3 APHINITY trial. METHODS: In the APHINITY trial, 4805 patients were randomly assigned to receive chemotherapy plus trastuzumab with pertuzumab or chemotherapy plus trastuzumab with placebo. Median follow-up was 74·1 months (IQR 68·3-75·4). We analysed 4262 haematoxylin and eosin-stained images using manual assessment, an automated digital approach, AI-based lymphocyte quantification (AI percentage lymphocytes), and two AI-derived spatial features (AI-TIL and immune hotspot). Interobserver reproducibility was assessed in 262 randomly chosen tumour samples scored independently by five pathologists. Multivariable Cox models were used to assess associations between TIL levels and invasive disease-free survival (primary outcome in APHINITY), distant recurrence-free interval, and overall survival. The heterogeneity of pertuzumab benefit was evaluated using subgroup analyses, subpopulation treatment effect pattern plot analyses, and nested Cox models with treatment-by-biomarker interaction terms. FINDINGS: Manual scoring showed high interobserver reproducibility (intraclass correlation coefficient 0·84 [95% CI 0·79-0·88]). Concordance between manual and automated methods was modest. AI-based scoring (AI percentage lymphocytes) reclassified 120 (11·6%) of 1035 node-positive tumours from immune-low (by manual scoring) to immune-high; this subgroup of patients showed greater separation of 5-year invasive disease-free survival curves between pertuzumab and placebo groups compared with patients whose tumours were concordantly classified as immune-low by both manual and AI-based approaches. Higher levels of TILs were associated with improved invasive disease-free survival for all sTIL measurement approaches and spatial measurements (hazard ratios [HRs] 0·41-0·93). Pertuzumab was associated with improved invasive disease-free survival at higher sTIL levels across all measurement approaches (HRs 0·36-0·48), but was not associated with higher values of spatial measures. The largest 6-year absolute improvements with pertuzumab were observed in patients with node-positive disease whose tumours scored in the highest level of immune infiltration of manual sTIL scoring (≥70·0%; mean absolute improvement 12·1 percentage points [SD 2·8]). In nested prognostic and predictive models, AI-based immune hotspot scores provided the most consistent additional information when combined with any sTIL measurement (all p<0·010). INTERPRETATION: Standardised manual sTIL scoring was reproducible, and digital and AI-based methods showed consistent prognostic stratification and potential for treatment-benefit stratification despite only modest correlation between platforms. AI spatial metrics provided complementary information beyond sTIL density and could support more scalable immune assessment. Future studies are needed to validate these approaches in independent cohorts and to clarify their clinical utility for stratifying contemporary HER2-directed therapies. FUNDING: None.

Evaluating AI and Manual Stromal Immune Cell Quantification in HER2 Positive Breast CancerAI tools help predict outcomes for HER2-positive breast cancer

How this fits prior evidence

Study Details

First-line immune checkpoint inhibitor combinations improve overall survival in extensive-stage small-cell lung cancer

Immune checkpoint inhibitors combined with chemotherapy improve small cell lung cancer survival

Clinical research that matters. Delivered to your inbox.

Evaluating AI and Manual Stromal Immune Cell Quantification in HER2 Positive Breast CancerAI tools help predict outcomes for HER2-positive breast cancer

How this fits prior evidence

More on Breast Cancer

Study Details

First-line immune checkpoint inhibitor combinations improve overall survival in extensive-stage small-cell lung cancer

Immune checkpoint inhibitors combined with chemotherapy improve small cell lung cancer survival

Clinical research that matters. Delivered to your inbox.

Related in Oncology

From Other Specialties