Skip to main content
Download PDF

Exploration of the relationship between tumor-infiltrating lymphocyte score and histological grade in breast cancer

BMC Cancer volume 24, Article number: 318 (2024) Cite this article

Abstract

Background

The histological grade is an important factor in the prognosis of invasive breast cancer and is vital to accurately identify the histological grade and reclassify of Grade2 status in breast cancer patients.

Methods

In this study, data were collected from 556 invasive breast cancer patients, and then randomly divided into training cohort (n = 335) and validation cohort (n = 221). All patients were divided into actual low risk group (Grade1) and high risk group (Grade2/3) based on traditional histological grade, and tumor-infiltrating lymphocyte score (TILs-score) obtained from multiphoton images, and the TILs assessment method proposed by International Immuno-Oncology Biomarker Working Group (TILs-WG) were also used to differentiate between high risk group and low risk group of histological grade in patients with invasive breast cancer. Furthermore, TILs-score was used to reclassify Grade2 (G2) into G2 /Low risk and G2/High risk. The coefficients for each TILs in the training cohort were retrieved using ridge regression and TILs-score was created based on the coefficients of the three kinds of TILs.

Results

Statistical analysis shows that TILs-score is significantly correlated with histological grade, and is an independent predictor of histological grade (odds ratio [OR], 2.548; 95%CI, 1.648–3.941; P < 0.0001), but TILs-WG is not an independent predictive factor for grade (P > 0.05 in the univariate analysis). Moreover, the risk of G2/High risk group is higher than that of G2/Low risk group, and the survival rate of patients with G2/Low risk is similar to that of Grade1, while the survival rate of patients with G2/High risk is even worse than that of patients with G3.

Conclusion

Our results suggest that TILs-score can be used to predict the histological grade of breast cancer and potentially to guide the therapeutic management of breast cancer patients.

Peer Review reports

Introduction

The histological grade of breast cancer is a well-recognized clinical variable, and the Nottingham grading system, modified by Elston and Ellis of Broome, is the most widely adopted grading classification system [1, 2]. The histological grade obtained from each slide is based on the degree of tubule formation, nuclear pleomorphism and mitotic counts. A numerical assessment of each criterion was carried out to classify the histological grade into 3 levels, namely, low grade (Grade1), intermediate grade (Grade2), and high grade (Grade3) [3]. Grade1 tumors are well-differentiated, have a low risk of recurrence and can be treated more conservatively; Grade2 tumors are moderately differentiated; and Grade 3 tumors are poorly differentiated, associated with a poor prognosis and should be treated more aggressively [1, 4]. Clinically, about half of breast cancer is classified as Grade1 or Grade3 status, but a considerable proportion is classified as Grade2 (30 − 60%). The concordance of breast cancer grading by pathologists shows that high consistency can be observed in Grade1 (Kappa value: 0.51) and Grade3 (Kappa value: 0.60), while low consistency was observed with Grade 2 (Kappa value: 0.33). Therefore, Grade2 cannot be used as a basis for clinical decision-making due to the intermediate risk of recurrence and low consistency [5, 6]. Furthermore, Engstrøm et al. reclassified all cases of breast cancer into six subtypes by gene expression analysis, and found that differences in breast cancer specific survival according to the subtypes occurred almost exclusively amongst patients with Grade2 tumors [7]. Previous studies have shown that histological grade, tumor size, degree of axillary lymph node (LN) involvement, age, hormone receptor status, HER2/neu status and the presence of lymphovascular invasion (LVI) were the prognostic factors for breast cancer [8]. The histological grade that considers both morphology and proliferation has unique prognostic significance compared to other prognostic factors and was widely used in clinical decision-making [4]. In addition, the histological grade has been incorporated into a variety of validated prognostic algorithms to determine the treatment of breast cancer, such as the Nottingham Prognostic Index, Adjuvant! Online and St Gallen guidelines [9,10,11]. Therefore, accurate identification of the histological grade of invasive breast cancer and refinement of Grade2 status have high clinical implications.

In recent decades, an increasing number of studies have focused on the prognostic impact of tumor microenvironment (TME), especially tumor-infiltrating lymphocytes (TILs) has been widely investigated as a prognostic and predictive biomarker in breast cancer [12,13,14]. Miyoshi et al. showed that for the ER+/HER2- (ER+: estrogen receptor positive) breast cancer, high levels of TILs predicted a shorter survival time after recurrence, and that the proportion of TILs was significantly correlated with histological grade [15]. Another study showed that increased TILs was associated with an excellent prognosis in node-positive, ER-/HER2- (ER-: estrogen receptor negative) breast cancer and that TILs was associated with high histological grade [16]. Many similar studies showed that TILs were associated with the histological grade of tumors [17,18,19]. In our previous study, the percentage of TILs, which is based on the consensus recommendation proposed by International Immuno-Oncology Biomarker Working Group (TILs-WG), was obtained from the hematoxylin and eosin (H&E)-stained sections to assess its prognostic value [20], but have not been utilized to investigate its relationship with histological grade and TILs.

Since its development in 1990, multiphoton microscopy (MPM) combining second harmonic generation (SHG) and two-photon excitation fluorescence (TPEF) signals has become an important imaging modality in biomedical sciences [21]. It can be applied to the non-invasive study of biological samples to obtain three-dimensional imaging with sub-micron resolution [22]. TPEF signals can be detected from endogenous fluorophores such as nicotinamide adenine dinucleotide (NADH), flavin adenine dinucleotide (FAD), and porphyrin, while SHG signals can be detected from non-centrosymmetric molecules such as collagen. Thus, MPM is able to simultaneously image cellular and extracellular matrix structures for label-free analysis of tissue samples [23]. It is considered one of the best non-invasive means of performing bioimaging in tissues and live animals, with the following advantages: (1) since multiphoton absorption occurs only at the focal point of the objective, MPM provides optical sectioning capability and avoids out-of-focus light bleaching; (2) the use of near-infrared (NIR) wavelengths allows for low scattering and deep tissue imaging [24]. The TPEF and SHG signals provide enhanced contrast and facilitate image interpretation, therefore, MPM has been widely developed and applied in the biomedical science field with the development of interdisciplinary medicine [25, 26].

In our previous study, MPM was used to image TILs in the TME of breast cancer and obtained a tumor-infiltrating lymphocyte score (TILs-score) for each patient, and statistical analysis showed that TILs-score was an independent prognostic factor for breast cancer [27]. We also improved the prognostic value by combining the TILs-score and TILs-WG in breast cancer [20]. In this work, we further investigated the relationship between TILs-WG, TILs-score and histological grade. The discrimination ability of the TILs-score was analyzed by the area under the curve (AUC), and a nomogram model combining the TILs-score with the clinical factors was developed for personalized prediction of histological grade in breast cancer patients.

Materials and methods

Study population and sample preparation

This retrospective study was approved by the Institutional Review Boards of Fujian Medical University Union Hospital. A total of 600 patients who were diagnosed with invasive breast cancer participated in our study. 44 patients were excluded due to neoadjuvant chemotherapy and radiotherapy, sections without lymphocytes, tumors located in lymph nodes, no available histological grade information and pathological reports, and 556 patients who passed quality control were included in this study and randomly segregated into the training cohort (n = 335) and validation cohort (n = 221), as shown in Fig. 1A. Tumor histological grade was retrospectively assessed according to the Nottingham system and Elston-Ellis grading method [1, 2], and is also treated as an ordinal categorical variable (Grade1/low grade, 13.49%; Grade2/intermediate grade, 51.08%; Grade3/high grade, 35.43%). In this study, all patients were divided into actual low risk group (Grade1) and high risk group (Grade2/3) based on traditional histological grade. The clinicopathologic characteristics of patients are shown in Table 1. Two continuous slices (5 μm thickness) were cut from each formalin-fixed paraffin-embedded (FFPE) tissue block. One slice was stained by H&E, and the other was used for MPM imaging.

Fig. 1
figure 1

A flowchart to display the selection of patients A and experimental scheme B

Full size image
Table 1 Characteristics of patients with breast cancer in actual low and high risk groups
Full size table

Multiphoton imaging system and data acquisition

The imaging system has been described in detail in the previous publication [28]. In short, it is based on a commercially upright laser scanning microscope (LSM 880, Zeiss, Germany) combined with a mode-locked femtosecond Ti: sapphire laser (Chameleon Ultra, Coherent, USA), tunable from 690 to 1064 nm. The excitation wavelength was set to 810 nm for all experiments, the SHG signal was collected from 395 to 415 nm (green color) by a GaAsP PMT, and the TPEF signal was collected from 428 to 695 nm (red color) by a 32-channel GaAsP PMT array detector. A Plan-Apochromat ×20 objective (NA = 0.8, Zeiss, Germany) was used to obtain high-resolution imaging.

The method of quantifying TILs-WG and TILs-score have been described in detail in previous studies [20, 27]. Simply put, the percentage of TILs in breast cancer was assessed separately by two pathologists on H&E-stained sections according to the standard method proposed by the International Immuno-Oncology Biomarker Working Group, and then the average percentage was taken as the final TILs percentage (TILs-WG) for each patient [20]. MPM can identify cells (such as tumor cells and lymphocytes) and extracellular matrix structures (such as collagen fibers) by endogenous signals. In the study, we first obtained the MPM images of each sample based on the 7–25 non-overlapping regions of interest (ROIs) marked on the H&E images within the tumor nest, tumor boundary, and invasive front, and one ROI may have multiple types of TILs, and a kind of TILs could exist in multiple ROIs. Then we observed the relative spatial positions of tumor cells, TILs, and collagen fibers, and classified the TILs in the TME into three patterns, namely TILs1-3, and recorded their occurrence frequencies separately. As shown in Supplementary Fig. 1, TILs-1 is defined as a pattern of infiltrating lymphocytes surrounded by tumor cells, TILs-2 is defined as a pattern of infiltrating lymphocytes around tumor cells, and TILs-3 is defined as infiltrating lymphocytes distributed in the TME without direct contact with tumor cells. Finally, we retrieved the coefficients for each TILs using ridge regression with cross validation based on the occurrence frequency of TILs1-3 in the training cohort, and fixed the coefficients of three kinds of TILs in a formula to calculate a patient-specific TILs score, which was used to study the relationship between TILs and the grade of breast cancer as shown in Fig. 1B.

Statistical analysis

The relationship between traditional clinical risk factors, TILs-score and histological grade (Grade1 group and Grade2/3 group) was explored using univariate and multivariate logistic regression analyses. A clinical (CLI) model based on the four clinicopathological factors (age, molecular subtype, tumor size, nodes metastasis) was developed and used to predict histological grade. The discrimination was measured by the AUC of the receiver operating characteristics (ROC) curve. A nomogram was created in the software package R using the nomogram function from the rms library. The nomogram created in the training cohort was applied to the validation cohort and the regression coefficients of the variables in the multiple regression were scaled to a score of 0-100. The variable with the largest regression coefficient had the greatest impact and was assigned a score of 100. The scores for each variable were summed to give a total score which was transformed into predicted probabilities. The calibration curve was drawn to evaluate the predictive ability of the nomogram model. The optimal cutoff value calculated from the training cohort was used to classify patients into low risk group and high risk group, and determine the sensitivity (SEN), specificity (SPE), positive predictive value (PPV), and negative predictive value (NPV) of the models, which was also applied to the validation cohort. The TILs-score was analyzed with Mann-Whitney U test, and differences between categorical variables were compared using the χ2 test. In order to estimate the association between histological grade and patients’ survival time, a Kaplan-Meier survival analysis was conducted and compared by log-rank test. The above statistical analyses were carried out using R 3.6.3, IBM SPSS Statistics 25 and Graph-Pad Prism 6.0.

Results

Clinicopathologic characteristics of patients

We included 556 patients with a median age of 49 years old (range, 24–84 years old). The clinicopathologic characteristics of training and validation cohorts are shown in Supplementary Table 1. There are no significant differences between the two cohorts, as well as the distribution of TILs-WG and TILs-score (P > 0.05). In addition, the molecular subtype, tumor size, progesterone receptor (PR) and TILs-score between the low risk and high risk groups are significantly different in both the training and validation cohorts (P < 0.05), as shown in Table 1.

Prediction of histological grade using TILs-score

We first analyzed the relationship between histological grade and patient DFS and the results were shown in Supplementary Fig. 2. We found that there is a difference in survival rates between Grade1 and Grade2 patients, the survival rates of patients with Grade1 are better than Grade2 (hazard ratio [HR], 1.935; 95% confidence interval [CI], 0.9467 to 3.167; P = 0.0759). Similarly, survival rates are different between Grade1 and Grade3 patients, patients with Grade1 have better survival rates than Grade3 (HR, 1.97; 95% CI, 0.9438 to 3.383; P = 0.0755) in the training cohort, although the differences are not statistically significant, but there is almost no difference in survival between Grade2 and Grade3 patients (HR, 1.013; 95% CI, 0.6633 to 1.548; P = 0.9415). While in the validation cohort, there is almost no difference between Grade1 and Grade2 (HR, 0.8386; P = 0.6222) and Grade3 (HR, 1.064; P = 0.865), and a slight difference in survival rates between Grade2 and Grade3 patients (HR, 1.28; P = 0.3276). Furthermore, a Kaplan-Meier survival analysis was performed on the whole cohort, and the results are consistent with that of the training cohort, with the differences between Grade1 and Grade2 (HR, 1.329; 95% CI, 0.8167 to 2.073; P = 0.2695) and Grade3 (HR, 1.478; 95% CI, 0.8911 to 2.299; P = 0.1391), and no difference between Grade2 and Grade3 (HR, 1.116; 95% CI, 0.8079 to 1.548; P = 0.5018), despite all of these differences have not statistically significance. The HR values, 95% CI, and P values are summarized in Supplementary Table 2. Therefore, all patients are divided into actual low risk group (Grade1) and high risk group (Grade2/3) based on traditional histological grade.

Table 2 Univariate and multivariate logistic regression analyses of the association of variables with pathologic grades in the training cohort
Full size table

The correlations of histological grade with clinical factors (age, molecular subtype, tumor size, nodes metastasis), expression of receptors or proteins on the surface of breast cancer cells (ER; PR; HER2), TILs-WG and TILs-score in the training cohort were assessed by the logistic regression analysis. Table 2 shows the results of the univariate and multivariate logistic regression analyses. The univariate logistic regression analysis reveals a significant association of histological grade with molecular subtype, tumor size, ER, PR and TILs-score (P < 0.05), while the correlation between TILs-WG and histologic grade is not statistically significant (odds ratio [OR], 1.027; 95% CI, 0.997 to 1.058; P = 0.078). After incorporating the factors that are significantly correlated with grade in the univariate logistic regression analysis into the multivariate analysis, the results show that TILs-score proves to be a strong independent predictor of histological grade (OR, 2.577; 95% CI, 1.696 to 3.917; P < 0.0001). To further evaluate the predictive accuracy of TILs-score, we estimated the AUC of TILs-score by the ROC analysis, and a CLI model which combines age, molecular subtype, tumor size, nodes metastasis was also developed to compare. As shown in Fig. 2A, the CLI model shows an AUC of 0.746 (95% CI, 0.696 to 0.792) and TILs-score shows an AUC of 0.747 (95% CI, 0.697 to 0.793) for predicting the low and high risk groups in the training cohort. The results indicate that the predictive ability of TILs-score is equivalent to that of the CLI model with four factors. What’s more, the AUC of TILs-score (AUC, 0.752; 95% CI, 0.689 to 0.807) is higher than that of CLI model (AUC, 0.694; 95% CI, 0.628 to 0.754) in the validation cohort (Fig. 2B). While the predictive ability of TILs-WG is not better than that of TILs-score and CLI model in both the training and validation cohorts (Training cohort: AUC = 0.613, 95% CI, 0.559 to 0.665; validation cohort: AUC = 0.557, 95% CI, 0.489 to 0.623). Since TILs-WG is not an independent predictive factor for histological grade (P > 0.05), we will not analyze the relationship between TILs-WG and grade in the following analysis.

Fig. 2
figure 2

The ROC curves of the CLI, TILs-WG, TILs-score and nomogram models in the training A and validation cohorts B

Full size image
Fig. 3
figure 3

A The nomogram combining the TILs-score with clinical factors in the training cohort. B The calibration curves of the nomogram in the training cohort. C The calibration curves of the nomogram in the validation cohort

Full size image

Furthermore, a nomogram which combines TILs-score with clinical factors is built for individualized prediction of low and high risk groups, and can increase the AUC of CLI model from 0.746 (95% CI, 0.696 to 0.792) to 0.802 (95% CI, 0.755 to 0.843) in the training cohort, and from 0.694 (95% CI, 0.628 to 0.754) to 0.776 (95% CI, 0.715 to 0.829) in the validation cohort, respectively. As displayed in Fig. 3, TILs-score has the greatest contribution to the prediction of histological grade. Several subgroup analyses are also conducted to assess the predictive performance of TILs-score. As shown in Table 3, TILs-score performs well except for HER2-enriched, triple-negative, and ER-negative patients. This is likely because TILs-score is more appropriate for assessing patients with ER-positive breast cancer rather than ER-negative breast cancer, as demonstrated by our previous works [20, 27].

Table 3 Prediction of clinicopathologically classified patients by the TILs-score
Full size table

Performance comparison of different prediction models

All patients are divided into the predicted low risk group and high risk group according to the optimal cutoff values obtained by the Youden index criterion from the training cohort. As shown in Table 1, the distribution of TILs-score between the actual low risk and high risk groups is significantly different in both the training and validation cohorts (P < 0.0001), and the median of TILs-score in the high risk is higher than that low risk group in both the training cohort (2.194 vs. 1.335) and validation cohort (2.219 vs. 1.428). In contrast, there is no significant difference in the distribution of TILs-WG in the actual low risk and high risk groups. Venn software is used to identify the overlapping patients between the actual and predicted, as shown in Fig. 4 and Supplementary Fig. 3. For example, in the training cohort, the actual number of patients diagnosed with low risk (Grade 1) is 47 and the predicted low risk of the CLI model is 121, of which 36 patients are actual low risk patients (Fig. 4A, left panel). While in the high risk (Grade 2/3) group, the overlapping number of patients between the actual and predicted is 203 (Fig. 4B, left panel). In both the training and validation cohorts, the number of overlapping patients of the TILs-score is comparable to that of the CLI model in both low risk and high risk groups (Fig. 4 and Supplementary Fig. 3, middle panel). The results show that TILs-score has a better predictive efficacy in the high risk group.

Fig. 4
figure 4

A Venn software was used to identify the overlapping patients between the actual and predicted in the low risk group of training cohort. B Venn software was used to identify the overlapping patients between the actual and predicted in the high risk group of training cohort

Full size image

As shown in Table 4, the SEN, SPE, PPV and NPV of the CLI model are 70.49% (95% CI, 64.9 to 75.7), 76.60% (95% CI, 62.0 to 87.7), 94.86% (95% CI, 91.6 to 96.9) and 29.75% (95% CI, 25.0 to 35.0) in the training cohort, respectively. The SEN, SPE, PPV and NPV of the TILs-score are 79.17% (95% CI, 74.0 to 83.7), 63.83% (95% CI, 48.5 to 77.3), 93.06% (95% CI, 90.1 to 95.2) and 33.33% (95%CI, 26.8 to 40.6) in the training cohort, respectively. It can be found that the predictive performance of the TILs-score is comparable to that of the CLI model. The AUC, SPE and PPV of the nomogram model are higher than those of the CLI model and TILs-score (Table 4). The same analyses are performed in the validation cohort and similar results are observed (Supplementary Table 3).

Table 4 Performance comparison of different models for predicting pathological grades in the training cohort
Full size table

Reclassification of Grade2 using TILs-score

We find that there are no differences in survival between Grade2 and Grade3 in the training cohort and in the whole cohort, and no difference in survival between Grade1 and Grade2 in the validation cohort. It appears that some patients with Grade2 have the same survival rate as patients with Grade1, while others have the same survival rate as patients with Grade3, as shown in Supplementary Fig. 2. Therefore, we tried to reclassify the patients into low risk and high risk groups using TILs-score, and as shown in Fig. 5A, it is found that patients in the high risk group have lower survival rate than patients in the low risk group. We also used TILs-score to reclassify patients with Grade2 into Grade2/High risk (G2/High risk) and Grade2/Low risk (G2/Low risk) and the survival rate of patients with Grade2 is significantly different (Fig. 5B). In the training cohort, patients in G2/High risk group have 4.09 fold higher risks for an event than patients in G2/Low risk group (HR, 4.09; 95% CI, 4.019 to 19.1; P < 0.0001), and the difference is statistically significant. Similarly, in the validation cohort and the whole cohort, patients in G2/High risk group have worse outcomes than patients in G2/Low risk group. In addition, we further used Kaplan-Meier survival curves to examine the correlation between G2/High risk group (G2/Low risk group) and clinical histological grade (Grade1, Grade2, Grade3). As shown in Fig. 5C, there is no significant difference in survival between G2/Low risk group and Grade1 in the training cohort (HR, 1.326; 95% CI, 0.6317 to 2.686; P = 0.4751), in the validation cohort (HR, 0.7291; 95% CI, 0.3203 to 1.571; P = 0.3983) and in the whole cohort (HR, 1.004; 95% CI, 0.5890 to 1.710; P = 0.9895). The survival rate of patients with G2/High risk is similar to Grade3 in validation cohort, and no significant difference in survival is found (HR, 1.153; 95% CI, 0.5642 to 2.386; P = 0.6871), but the survival rate of patients with G2/High risk is lower than that of patients with Grade3 in both the training cohort and the whole cohort (Training cohort: HR, 2.933; 95% CI, 2.237 to 9.042; P < 0.0001; Whole cohort: HR, 1.964; 95% CI, 1.378 to 3.723; P = 0.0014). The HR and P values of the three cohorts are shown in Table 5.

Fig. 5
figure 5

A Kaplan-Meier curves for the patients re-stratified into low and high risk group by TILs-score in the three cohorts. B Kaplan-Meier curves for the re-stratified G2 cases by TILs-score. C Kaplan-Meier curves comparing the G2/High-risk and G2/Low-risk groups with clinical histological grade in the three cohorts

Full size image
Table 5 Hazard ratios for 5-year DFS between different groups in the three cohorts
Full size table

Discussion

Breast cancer is a complex and heterogeneous disease as reflected by different morphological presentations, molecular features and responses to treatment. The routine clinical management of breast cancer relies on the availability of clinicopathological prognostic and predictive factors to support clinicians in their decision-making. Studies have shown that the main determinants of survival in breast cancer are tumor size, lymph node status and histological grade. Thereinto, histological grade represents a morphological assessment of the biological characteristics of tumor and has been shown to provide important information related to the clinical behaviors of breast cancer [29, 30]. The higher histological grade is associated with the lower overall survival and long-term survival [31, 32]. Galimberti et al. stated that histological grade was a significant predictor of disease-free survival [33], and Rakha et al. showed that histological grade was an independent prognostic factor in invasive lobular carcinoma of breast cancer [34]. In addition to the accurate assessment of histopathological features on H&E staining by the morphology of cells, imaging techniques such as X-ray mammography [35], magnetic resonance imaging (MRI) [36], and ultrasonography [37] can be used to predict histopathological features of breast cancer, but the resolution of these imaging methods is still limited. Currently, classification prediction for histological grade using artificial intelligence is also available, but the performance is affected by poorer consistency of tumors in the intermediate grade (Grade2) [38]. At present, the histological grade of tumors mainly refers to a semi-quantitative evaluation of the morphological characteristics of the tumor cells.

Many studies have shown that the microenvironment surrounding the tumor seed plays a crucial role in tumor growth and development [12]. TILs as an important component of TME has been shown to be an independent prognostic factor in breast cancer [16, 17]. Our previous study revealed that TILs-score obtained by MPM was an independent prognostic factor of breast cancer [27], and in this study, we further confirm the relationship between the TILs-score and the histological grade, and higher TILs-score is associated with a higher grade. Multivariate logistic regression analysis shows that TILs-score is an independent predictor of histological grade (OR, 2.577; P < 0.0001) and has comparable predictive power to the CLI model combining four factors (training cohort, AUC, 0.747 vs. 0.746; validation cohort, AUC, 0.752 vs. 0.694). Desmedt et al. demonstrated that only histological grade and proliferation modules correlated with relapse-free survival (RFS) in ER-positive/HER2-negative breast cancer [39]. In LN-negative [40, 41] or LN-positive [42, 43] breast cancer irrespective of ER expression and patients with ER-positive breast cancer without [44] or with neoadjuvant endocrine therapy [45], histological grade was an independent prognostic factor. Additionally, histological grade could complement the LN stage as it can influence the outcome of patients in different LN stage categories [43]. TILs-core was further used to divide patients into the low risk and high risk groups based on the optimal cutoff values, and over 90% of patients were predicted to be high risk who were truly high risk (training cohort, PPV, 93.06%, 95%CI, 90.1 to 95.2; validation cohort, PPV, 92.80%, 95%CI, 89.0 to 95.4). More importantly, the nomogram model which combined the TILs-score with clinical factors would increase the PPV of the CLI model from 94.86% (95% CI, 91.6 to 96.9) to 96.72% (95% CI, 93.3 to 98.4) in the training cohort and from 93.6% (95% CI, 89.5 to 96.2) to 96.0% (95% CI, 91.6 to 98.2) in the validation cohort, respectively. The results show a better predictive performance in the high risk group. The reason for the low NPV of all models may be because the number of patients actually diagnosed as low risk in the dataset is too small to provide a relevant prediction.

A study has shown that high-grade tumors carry a risk of early recurrence and death, and require consideration of timely use of adjuvant chemotherapy, while patients with low-grade tumors are almost invariably ER-positive and can be offered long-term follow-up with or without potentially less toxic systemic therapies. (i.e. endocrine therapy) [29]. However, up to half of breast cancer cases are categorized as “Grade2” in routine clinical practice, which may include a number of low grade (Grade1) and high grade (Grade2) tumors, so Grade2 is judged to be an intermediate risk group with limited clinical value [46]. As a result, many researchers opened several investigations at the patients with Grade2, for example, Sotiriou et al. used genes to investigate histological grade and found that gene expression grade indices correlate with Grades1 and Grade3, whereas the index for Grade2 spans the values of Grade1 and Grade3, and patients with Grade2 can be reclassified into low and high risk group via the index to improve the prognostic value [5]; similarly, Ivshina et al. used gene expression signatures for classification prediction of grade and the results showed that they could accurately classify Grade1 and Grade3, and could classify Grade2 tumors into two highly distinguishable categories (Grade2a and Grade2b genetic grades) whose survival outcomes were highly similar to those of Grade1 and Grade3, respectively [47]; Li et al. developed a qualitative transcriptional signature using within-sample relative expression orderings of gene pairs to redefine Grade1 and Grade3, and Grade2 is further reclassified into the redefined Grade1 and Grade3 [48]; Wang et al. reclassified the Nottingham histological grade 2 (NHG2) into DeepGrade2-low (DG2-low) and DG2-high groups, and found that DG2-high had an increased risk of recurrence compared to DG2-low, and besides DG2-low also had a similar phenotype to NHG1 and DG2-high with NHG3 [4]. In this work, TILs-score was used to reclassify G2 into G2 /Low risk and G2/High risk, and the results showed that TILs-score can classify G2 tumors into two highly distinguishable categories, and the risk of G2/High risk group is higher than that of G2/Low risk group. Moreover, the survival rate of patients with G2/Low risk is similar to that of patients with pathologically-determined histological grade1 (G1), while the survival rate of patients with G2/High risk is worse than that of patients with G3. Our findings are similar to those of previous studies, suggesting that patients with G2 should be re-stratified to screen out the high risk from low risk patients, prevent some patients from undertreatment or overtreatment, reduce the side effects and costs of chemotherapy, and improve patient survival and the quality of life.

Conclusions

In summary, our results reveal that TILs-score obtained from MPM images is a simple method to differentiate the histological grade of breast cancer (Grade1 and Grade2/3). In addition, TILs-score could be used to re-stratify intermediate risk breast cancer patients (Grade2) to increase the information needed for clinical decision making, and thereby improve the survival rate and prognosis of breast cancer patients.

Data availability

Data are available from the corresponding author on reasonable request.

Abbreviations

LN:

lymph node

LVI:

lymphovascular invasion

TME:

tumor microenvironment

TILs:

tumor-infiltrating lymphocytes

H&E:

hematoxylin and eosin

MPM:

multiphoton microscopy

SHG:

second harmonic generation

TPEF:

two-photon excitation fluorescence

NADH:

nicotinamide adenine dinucleotide

FAD:

flavin adenine dinucleotide

NIR:

near-infrared

TILs-score:

tumor-infiltrating lymphocyte score

AUC:

area under the curve

FFPE:

formalin-fixed paraffin-embedded

ROI:

region of interest

CLI:

clinical

ROC:

receiver operating characteristics

SEN:

sensitivity

SPE:

specificity

PPV:

positive predictive value

NPV:

negative predictive value

DFS:

disease-free survival

HR:

hazard ratio

CI:

confidence interval

OR:

odds ratio

MRI:

magnetic resonance imaging

RFS:

relapse-free survival

References

  1. Elston CW, Ellis IO. Pathological prognostic factors in breast cancer. I. The value of histological grade in breast cancer: experience from a large study with long-term follow-up. Histopathology. 1991;19(5):403–10.

    Article  CAS  PubMed  Google Scholar 

  2. Jerevall PL, Ma XJ, Li H, Salunga R, Kesty NC, Erlander MG, et al. Prognostic utility of HOXB13: IL17BR and molecular grade index in early-stage breast cancer patients from the Stockholm trial. Brit J Cancer. 2011;104(11):1762–9.

    Article  CAS  PubMed Central  PubMed  Google Scholar 

  3. Theissig F, Kunze KD, Haroske G, Meyer W. Histological grading of breast cancer: interobserver, reproducibility and prognostic significance. Patholo Res Pract. 1990;186(6):732–6.

    Article  CAS  Google Scholar 

  4. Wang Y, Acs B, Robertson S, Liu B, Solorzano L, Wahlby C, et al. Improved breast cancer histological grading using deep learning. Ann Oncol. 2022;33(1):89–98.

    Article  CAS  PubMed  Google Scholar 

  5. Sotiriou C, Wirapati P, Loi S, Harris A, Fox S, Smeds J, et al. Gene expression profiling in breast cancer: understanding the molecular basis of histologic grade to improve prognosis. J Natl Cancer I. 2006;98(4):262–72.

    Article  CAS  Google Scholar 

  6. Rakha EA, Aleskandarani M, Toss MS, Green AR, Ball G, Ellis IO, et al. Breast cancer histologic grading using digital microscopy: concordance and outcome association. J Clin Pathol. 2018;71(8):680–6.

    Article  PubMed  Google Scholar 

  7. Engstrøm MJ, Opdahl S, Hagen AI, Romundstad PR, Akslen LA, Haugen OA, et al. Molecular subtypes, histopathological grade and survival in a historic cohort of breast cancer patients. Breast Cancer Res Tr. 2013;140(3):463–73.

    Article  Google Scholar 

  8. Wo JY, Chen K, Neville BA, Lin NU, Punglia RS. Effect of very small tumor size on cancer-specific mortality in node-positive breast cancer. J Clin Oncol. 2011;29(19):2619–27.

    Article  PubMed Central  PubMed  Google Scholar 

  9. Galea MH, Blamey RW, Elston CE, Ellis IO. The Nottingham Prognostic Index in primary breast cancer. Breast Cancer Res TR. 1992;22(3):207–19.

    Article  CAS  Google Scholar 

  10. Mook S, Schmidt MK, Rutgers EJ, van de Velde AO, Visser O, Rutgers SM, et al. Calibration and discriminatory accuracy of prognosis calculation for breast cancer with the online adjuvant! Program: a hospital-based retrospective cohort study. Lancet Oncol. 2009;10(11):1070–6.

    Article  PubMed  Google Scholar 

  11. Goldhirsch A, Ingle JN, Gelber RD, Coates AS, Thurlimann B, Senn HJ. Thresholds for therapies: highlights of the St Gallen International Expert Consensus on the primary therapy of early breast cancer. Ann Oncol. 2009;20(8):1319–29.

    Article  CAS  PubMed Central  PubMed  Google Scholar 

  12. Quail DF, Joyce JA. Microenvironmental regulation of tumor progression and metastasis. Nat Med. 2013;19(11):1423–37.

    Article  CAS  PubMed Central  PubMed  Google Scholar 

  13. Denkert C, von Minckwitz G, Darb-Esfahani S, Lederer B, Heppner BI, Weber KE, et al. Tumour-infiltrating lymphocytes and prognosis in different subtypes of breast cancer: a pooled analysis of 3771 patients treated with neoadjuvant therapy. Lancet Oncol. 2018;19(1):40–50.

    Article  PubMed  Google Scholar 

  14. Tekpli X, Lien T, Røssevold AH, Nebdal D, Borgen E, Ohnstad HO, et al. An independent poor-prognosis subtype of breast cancer defined by a distinct tumor immune microenvironment. Nat Commun. 2019;10:5499.

    Article  CAS  PubMed Central  PubMed  Google Scholar 

  15. Miyoshi Y, Shien T, Ogiya A, Ishida N, Yamazaki K, Horii R, et al. Associations in tumor infiltrating lymphocytes between clinicopathological factors and clinical outcomes in estrogen receptor–positive/human epidermal growth factor receptor type 2 negative breast cancer. Oncol Lett. 2019;17(2):2177–86.

    CAS  PubMed  Google Scholar 

  16. Loi S, Sirtaine N, Piette F, Salgado R, Viale G, Van Eenoo F, et al. Prognostic and predictive value of tumor-infiltrating lymphocytes in a phase III randomized adjuvant breast cancer trial in node-positive breast cancer comparing the addition of docetaxel to doxorubicin with doxorubicin-based chemotherapy: BIG 02–98. J Clin Oncol. 2013;31(7):860–7.

    Article  CAS  PubMed  Google Scholar 

  17. Loi S, Drubay D, Adams S, Pruneri G, Francis PA, Lacroix-Triki M, et al. Tumor-infiltrating lymphocytes and prognosis: a pooled individual patient analysis of early-stage triple-negative breast cancers. J Clin Oncol. 2019;37(7):559–69.

    Article  PubMed Central  PubMed  Google Scholar 

  18. Criscitiello C, Vingiani A, Maisonneuve P, Viale G, Vialel G, Curigliano G. Tumor-infiltrating lymphocytes (TILs) in ER+/HER2- breast cancer. Breast Cancer Res TR. 2020;183(2):347–54.

    Article  CAS  Google Scholar 

  19. Ali HR, Dariush A, Thomas J, Provenzano E, Dunn J, Hiller L, et al. Lymphocyte density determined by computational pathology validated as a predictor of response to neoadjuvant chemotherapy in breast cancer: secondary analysis of the ARTemis trial. Ann Oncol. 2017;28(8):1832–5.

    Article  CAS  PubMed Central  PubMed  Google Scholar 

  20. He J, Kang D, Xu M, Han Z, Guo W, Fu F et al. Combining the guidelines and multiphoton imaging methods to improve the prognostic value of tumor-infiltrating lymphocytes in breast cancer. J Biophotonics, 2023; e202300060.

  21. Denk W, Strickler JH, Webb WW. Two-photon laser scanning fluorescence microscopy. Science. 1990;248(4951):73–6.

    Article  ADS  CAS  PubMed  Google Scholar 

  22. Zhao Y, Sheng M, Huang L, Tang S. Design of a fiber-optic multiphoton microscopy handheld probe. Biomed Opt Express. 2016;7(9):3425–37.

    Article  PubMed Central  PubMed  Google Scholar 

  23. Huang L, Mills AK, Zhao Y, Jones DJ, Tang S. Miniature fiber-optic multiphoton microscopy system using frequency-doubled femtosecond Er-doped fiber laser. Biomed Opt Express. 2016;7(5):1948–56.

    Article  PubMed Central  PubMed  Google Scholar 

  24. Akhoundi F, Peyghambarian N. Single-cavity dual-wavelength all-fiber femtosecond laser for multimodal multiphoton microscopy. Biomed Opt Express. 2020;11(5):2761–7.

    Article  CAS  PubMed Central  PubMed  Google Scholar 

  25. Cui Y, Zhong Q, Sun D, Chen Y, Jiang Z, Yang X, et al. Evaluation of histopathological response to neoadjuvant therapy in rectal cancer using slide-free, stain-free multimodal multiphoton microscopy. J Biophotonics. 2022;15(10):e202200079.

    Article  CAS  PubMed  Google Scholar 

  26. Pavlova IP, Nair SS, Lundon D, Sobotka S, Roshandel R, Treacy PJ, et al. Tewari, Multiphoton microscopy for identifying collagen signatures associated with biochemical recurrence in prostate cancer patients. J Pers Med. 2021;11(11):1061.

    Article  PubMed Central  PubMed  Google Scholar 

  27. He J, Fu F, Wang W, Xi G, Guo W, Zheng L, et al. Prognostic value of tumour-infiltrating lymphocytes based on the evaluation of frequency in patients with oestrogen receptor-positive breast cancer. Eur J Cancer. 2021;154:217–26.

    Article  CAS  PubMed  Google Scholar 

  28. Xi G, Guo W, Kang D, Ma J, Fu F, Qiu L, et al. Large-scale tumor-associated collagen signatures identify high-risk breast cancer patients. Theranostics. 2021;11(7):3229–43.

    Article  CAS  PubMed Central  PubMed  Google Scholar 

  29. Rakha EA, Reis-Filho JS, Baehner F, Dabbs DJ, Decker T, Eusebi V, et al. Breast cancer prognostic classification in the molecular era: the role of histological grade. Breast Cancer Res. 2010;12(4):207.

    Article  PubMed Central  PubMed  Google Scholar 

  30. Wishart GC, Greenberg DC, Chou P, Brown CH, Duffy S, Purushotham AD. Treatment and survival in breast cancer in the Eastern Region of England. Ann Oncol. 2010;21(2):291–6.

    Article  CAS  PubMed  Google Scholar 

  31. Liedtke C, Hatzis C, Symmans WF, Desmedt C, Haibe-Kains B, Valero V, et al. Genomic grade index is associated with response to chemotherapy in patients with breast cancer. J Clin Oncol. 2009;27(19):3185–91.

    Article  PubMed Central  PubMed  Google Scholar 

  32. Soerjomataram I, Louwman MWJ, Ribot JG, Roukema JA, Coebergh JWW. An overview of prognostic factors for long-term survivors of breast cancer. Breast Cancer Res TR. 2008;107(3):309–30.

    Article  CAS  Google Scholar 

  33. Galimberti V, Cole BF, Zurrida S, Viale G, Luini A, Veronesi P, et al. Axillary dissection versus no axillary dissection in patients with sentinel-node micrometastases (IBCSG 23–01): a phase 3 randomised controlled trial. Lancet Oncol. 2013;14(4):297–305.

    Article  PubMed Central  PubMed  Google Scholar 

  34. Rakha EA, El-Sayed ME, Menon S, Green AR, Lee AHS, Ellis IO. Histologic grading is an independent prognostic factor in invasive lobular carcinoma of the breast. Breast Cancer Res TR. 2008;111(1):121–7.

    Article  Google Scholar 

  35. La Forgia D, Fanizzi A, Campobasso F, Bellotti R, Didonna V, Lorusso V, et al. Radiomic analysis in contrast-enhanced spectral mammography for predicting breast cancer histological outcome. Diagnostics. 2020;10(9):708.

    Article  PubMed Central  PubMed  Google Scholar 

  36. Fan M, Liu Z, Xie S, Xu M, Wang S, Gao X, et al. Integration of dynamic contrast-enhanced magnetic resonance imaging and T2-weighted imaging radiomic features by a canonical correlation analysis-based feature fusion method to predict histological grade in ductal breast carcinoma. Phy Med Biol. 2019;64(21):215001.

    Article  CAS  Google Scholar 

  37. Grajo JR, Barr RG. Strain elastography for prediction of breast cancer tumor grades. J Ultras Med. 2014;33(1):129–34.

    Article  Google Scholar 

  38. Lagree A, Shiner A, Alera MA, Fleshner L, Law E, Law B, et al. Assessment of digital pathology imaging biomarkers associated with breast cancer histologic grade. Curr Oncol. 2021;28(6):4298–316.

    Article  PubMed Central  PubMed  Google Scholar 

  39. Desmedt C, Haibe-Kains B, Wirapati P, Buyse M, Larsimont D, Bontempi G, et al. Biological processes associated with breast cancer clinical outcome depend on the molecular subtypes. Clin Cancer Res. 2008;14(16):5158–65.

    Article  CAS  PubMed  Google Scholar 

  40. Saimura M, Fukutomi T, Tsuda H, Sato H, Miyamoto K, Akashi-Tanaka S, et al. Prognosis of a series of 763 consecutive node-negative invasive breast cancer patients without adjuvant therapy: analysis of clinicopathological prognostic factor. J Surg Oncol. 1999;71(2):101–5.

    Article  CAS  PubMed  Google Scholar 

  41. Mirza AN, Mirza NQ, Vlastos G, Singletary SE. Prognostic factors in node-negative breast cancer: a review of studies with sample size more than 200 and follow-up more than 5 years. Ann Surg. 2002;235(1):10–26.

    Article  PubMed Central  PubMed  Google Scholar 

  42. Simpson JF, Gray R, Dressler LG, Cobau CD, Falkson CI, Gilchrist KW, et al. Prognostic value of histologic grade and proliferative activity in axillary node-positive breast cancer: results from the eastern cooperative oncology group companion study, est 4189. J Clin Oncol. 2000;18(10):2059–69.

    Article  CAS  PubMed  Google Scholar 

  43. Rakha EA, El-Sayed ME, Lee AHS, Elston CW, Grainge MJ, Hodi Z, et al. Prognostic significance of Nottingham histologic grade in invasive breast carcinoma. J Clin Oncol. 2008;26(19):3153–8.

    Article  PubMed  Google Scholar 

  44. Wirapati P, Sotiriou C, Kunkel S, Farmer P, Pradervand S, Haibe-Kains B, et al. Meta-analysis of gene expression profiles in breast cancer: toward a unified understanding of breast cancer subtyping and prognosis signatures. Breast Cancer Res. 2008;10(4):R65.

    Article  PubMed Central  PubMed  Google Scholar 

  45. Ellis MJ, Tao Y, Luo J, A’Hern R, Evans DB, Bhatnagar AS, et al. Outcome prediction for estrogen receptor-positive breast cancer based on postneoadjuvant endocrine therapy tumor characteristics. J Natl Cancer I. 2008;100(19):1380–8.

    Article  CAS  Google Scholar 

  46. Jaroensri R, Wulczyn E, Hegde N, Brown T, Flament-Auvigne I, Tan F, et al. Deep learning models for histologic grading of breast cancer and association with disease prognosis. npj Breast Cancer. 2022;8(1):113.

    Article  PubMed Central  PubMed  Google Scholar 

  47. Ivshina AV, George J, Senko O, Mow B, Putti TC, Smeds J, et al. Genetic reclassification of histologic grade delineates new clinical subtypes of breast cancer. Cancer Res. 2006;66(21):10292–301.

    Article  CAS  PubMed  Google Scholar 

  48. Li J, Jiang W, Liang Q, Liu G, Dai Y, Zheng H, et al. A qualitative transcriptional signature to reclassify histological grade of ER-positive breast cancer patients. BMC Genomics. 2020;21(1):283.

    Article  CAS  PubMed Central  PubMed  Google Scholar 

Download references

Funding

This work was supported by the National Natural Science Foundation of China (Grant Nos. 82171991, 82172800), and Natural Science Foundation of Fujian Province (Nos. 2023J01082, 2020J011008, 2022J01170, 2020J01154, 2023J011125).

Author information

Author notes

  1. Deyong Kang, Chuan Wang and Zhonghua Han have contributed equally to this work.

Authors and Affiliations

  1. Department of Pathology, Fujian Medical University Union Hospital, 350001, Fuzhou, P. R. China

    Deyong Kang

  2. Breast Surgery Ward, Department of General Surgery, Fujian Medical University Union Hospital, 350001, Fuzhou, P. R. China

    Chuan Wang, Zhonghua Han, Wenhui Guo & Fangmeng Fu

  3. Key Laboratory of OptoElectronic Science and Technology for Medicine of Ministry of Education, Fujian Provincial Key Laboratory of Photonics Technology, College of Photonic and Electronic Engineering, Fujian Normal University, 350007, Fuzhou, P. R. China

    Liqin Zheng, Xiahui Han, Lianhuang Li & Jianxin Chen

  4. College of Physics and Electronic Information Engineering, Minjiang University, 350108, Fuzhou, P. R. China

    Lida Qiu

  5. School of Science, Jimei University, 361021, Xiamen, P. R. China

    Jiajia He

Authors
  1. Deyong Kang
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Chuan Wang
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Zhonghua Han
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Liqin Zheng
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Wenhui Guo
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Fangmeng Fu
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Lida Qiu
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Xiahui Han
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Jiajia He
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Lianhuang Li
    View author publications

    You can also search for this author in PubMed Google Scholar

  11. Jianxin Chen
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

DYK, JJH, LHL and JXC conceived the experiments and directed the research. DYK, ZHH, CW, WHG and FMF were responsible for sample collection and preparation. LQZ, XHH and JJH contributed to the acquisition of multiphoton images. DYK, LDQ, JJH and LHL interpreted the results and drafted the manuscript. All authors critically reviewed the article and approved the final submission.

Corresponding authors

Correspondence to Jiajia He, Lianhuang Li or Jianxin Chen.

Ethics declarations

Ethics approval and consent to participate

The study was approved by the Institutional Review Board of Fujian Medical University Union Hospital (Approval Number: 2020KJT010). Due to the nature of retrospective study, the need for informed consent was waived by the Institutional Review Board of Fujian Medical University Union Hospital. All methods in this study were carried out in accordance with relevant guidelines and regulations.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary Material 1

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Kang, D., Wang, C., Han, Z. et al. Exploration of the relationship between tumor-infiltrating lymphocyte score and histological grade in breast cancer. BMC Cancer 24, 318 (2024). https://0-doi-org.brum.beds.ac.uk/10.1186/s12885-024-12069-0

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://0-doi-org.brum.beds.ac.uk/10.1186/s12885-024-12069-0

Keywords

BMC Cancer

ISSN: 1471-2407

Contact us