Retrospective validation study of an artificial neural network-based preoperative decision-support tool for noninvasive lymph node staging (NILS) in women with primary breast cancer (ISRCTN14341750).

BACKGROUND: Surgical sentinel lymph node biopsy (SLNB) is routinely used to reliably stage axillary lymph nodes in early breast cancer (BC). However, SLNB may be associated with postoperative arm morbidities. For most patients with BC undergoing SLNB, the findings are benign, and the procedure is currently questioned. A decision-support tool for the prediction of benign sentinel lymph nodes based on preoperatively available data has been developed using artificial neural network modelling. METHODS: This was a retrospective geographical and temporal validation study of the noninvasive lymph node staging (NILS) model, based on preoperatively available data from 586 women consecutively diagnosed with primary BC at two sites. Ten preoperative clinicopathological characteristics from each patient were entered into the web-based calculator, and the probability of benign lymph nodes was predicted. The performance of the NILS model was assessed in terms of discrimination with the area under the receiver operating characteristic curve (AUC) and calibration, that is, comparison of the observed and predicted event rates of benign axillary nodal status (N0) using calibration slope and intercept. The primary endpoint was axillary nodal status (discrimination, benign [N0] vs. metastatic axillary nodal status [N+]) determined by the NILS model compared to nodal status by definitive pathology. RESULTS: The mean age of the women in the cohort was 65 years, and most of them (93%) had luminal cancers. Approximately three-fourths of the patients had no metastases in SLNB (N0 74% and 73%, respectively). The AUC for the predicted probabilities for the whole cohort was 0.6741 (95% confidence interval: 0.6255-0.7227). More than one in four patients (n = 151, 26%) were identified as candidates for SLNB omission when applying the predefined cut-off for lymph node-negative status from the development cohort. The NILS model showed the best calibration in patients with a predicted high probability of healthy axilla. CONCLUSION: The performance of the NILS model was satisfactory. In approximately every fourth patient, SLNB could potentially be omitted. Considering the shift from postoperatively to preoperatively available predictors in this validation study, we have demonstrated the robustness of the NILS model. The clinical usability of the web interface will be evaluated before its clinical implementation. TRIAL REGISTRATION: Registered in the ISRCTN registry with study ID ISRCTN14341750. Date of registration 23/11/2018.

Noninvasive Staging of Lymph Node Status in Breast Cancer Using Machine Learning: External Validation and Further Model Development.

BACKGROUND: Most patients diagnosed with breast cancer present with a node-negative disease. Sentinel lymph node biopsy (SLNB) is routinely used for axillary staging, leaving patients with healthy axillary lymph nodes without therapeutic effects but at risk of morbidities from the intervention. Numerous studies have developed nodal status prediction models for noninvasive axillary staging using postoperative data or imaging features that are not part of the diagnostic workup. Lymphovascular invasion (LVI) is a top-ranked predictor of nodal metastasis; however, its preoperative assessment is challenging. OBJECTIVE: This paper aimed to externally validate a multilayer perceptron (MLP) model for noninvasive lymph node staging (NILS) in a large population-based cohort (n=18,633) and develop a new MLP in the same cohort. Data were extracted from the Swedish National Quality Register for Breast Cancer (NKBC, 2014-2017), comprising only routinely and preoperatively available documented clinicopathological variables. A secondary aim was to develop and validate an LVI MLP for imputation of missing LVI status to increase the preoperative feasibility of the original NILS model. METHODS: Three nonoverlapping cohorts were used for model development and validation. A total of 4 MLPs for nodal status and 1 LVI MLP were developed using 11 to 12 routinely available predictors. Three nodal status models were used to account for the different availabilities of LVI status in the cohorts and external validation in NKBC. The fourth nodal status model was developed for 80% (14,906/18,663) of NKBC cases and validated in the remaining 20% (3727/18,663). Three alternatives for imputation of LVI status were compared. The discriminatory capacity was evaluated using the validation area under the receiver operating characteristics curve (AUC) in 3 of the nodal status models. The clinical feasibility of the models was evaluated using calibration and decision curve analyses. RESULTS: External validation of the original NILS model was performed in NKBC (AUC 0.699, 95% CI 0.690-0.708) with good calibration and the potential of sparing 16% of patients with node-negative disease from SLNB. The LVI model was externally validated (AUC 0.747, 95% CI 0.694-0.799) with good calibration but did not improve the discriminatory performance of the nodal status models. A new nodal status model was developed in NKBC without information on LVI (AUC 0.709, 95% CI: 0.688-0.729), with excellent calibration in the holdout internal validation cohort, resulting in the potential omission of 24% of patients from unnecessary SLNBs. CONCLUSIONS: The NILS model was externally validated in NKBC, where the imputation of LVI status did not improve the model's discriminatory performance. A new nodal status model demonstrated the feasibility of using register data comprising only the variables available in the preoperative setting for NILS using machine learning. Future steps include ongoing preoperative validation of the NILS model and extending the model with, for example, mammography images.

Mammographic density in relation to breast cancer recurrence and survival in women receiving neoadjuvant chemotherapy.

Objective: The association between mammographic density (MD) and breast cancer (BC) recurrence and survival remains unclear. Patients receiving neoadjuvant chemotherapy (NACT) are in a vulnerable situation with the tumor within the breast during treatment. This study evaluated the association between MD and recurrence/survival in BC patients treated with NACT. Methods: Patients with BC treated with NACT in Sweden (2005-2016) were retrospectively included (N=302). Associations between MD (Breast Imaging-Reporting and Data System (BI-RADS) 5th Edition) and recurrence-free/BC-specific survival at follow-up (Q1 2022) were addressed. Hazard ratios (HRs) for recurrence/BC-specific survival (BI-RADS a/b/c vs. d) were estimated using Cox regression analysis and adjusted for age, estrogen receptor status, human epidermal growth factor receptor 2 status, axillary lymph node status, tumor size, and complete pathological response. Results: A total of 86 recurrences and 64 deaths were recorded. The adjusted models showed that patients with BI-RADS d vs. BI-RADS a/b/c had an increased risk of recurrence (HR 1.96 (95% confidence interval (CI) 0.98-3.92)) and an increased risk of BC-specific death (HR 2.94 (95% CI 1.43-6.06)). Conclusion: These findings raise questions regarding personalized follow-up for BC patients with extremely dense breasts (BI-RADS d) pre-NACT. More extensive studies are required to confirm our findings.

The implementation of NILS: A web-based artificial neural network decision support tool for noninvasive lymph node staging in breast cancer.

Objective: To implement artificial neural network (ANN) algorithms for noninvasive lymph node staging (NILS) to a decision support tool and facilitate the option to omit surgical axillary staging in breast cancer patients with low-risk of nodal metastasis. Methods: The NILS tool is a further development of an ANN prototype for the prediction of nodal status. Training and internal validation of the original algorithm included 15 clinical and tumor-related variables from a consecutive cohort of 800 breast cancer cases. The updated NILS tool included 10 top-ranked input variables from the original prototype. A workflow with four ANN pathways was additionally developed to allow different combinations of missing preoperative input values. Predictive performances were assessed by area under the receiver operating characteristics curves (AUC) and sensitivity/specificity values at defined cut-points. Clinical utility was presented by estimating possible sentinel lymph node biopsy (SLNB) reduction rates. The principles of user-centered design were applied to develop an interactive web-interface to predict the patient's probability of healthy lymph nodes. A technical validation of the interface was performed using data from 100 test patients selected to cover all combinations of missing histopathological input values. Results: ANN algorithms for the prediction of nodal status have been implemented into the web-based NILS tool for personalized, noninvasive nodal staging in breast cancer. The estimated probability of healthy lymph nodes using the interface showed a complete concordance with estimations from the reference algorithm except in two cases that had been wrongly included (ineligible for the technical validation). NILS predictive performance to distinguish node-negative from node-positive disease, also with missing values, displayed AUC ranged from 0.718 (95% CI, 0.687-0.748) to 0.735 (95% CI, 0.704-0.764), with good calibration. Sensitivity 90% and specificity 34% were demonstrated. The potential to abstain from axillary surgery was observed in 26% of patients using the NILS tool, acknowledging a false negative rate of 10%, which is clinically accepted for the standard SLNB technique. Conclusions: The implementation of NILS into a web-interface are expected to provide the health care with decision support and facilitate preoperative identification of patients who could be good candidates to avoid unnecessary surgical axillary staging.

The implementation of a noninvasive lymph node staging (NILS) preoperative prediction model is cost effective in primary breast cancer.

PURPOSE: The need for sentinel lymph node biopsy (SLNB) in clinically node-negative (cN0) patients is currently questioned. Our objective was to investigate the cost-effectiveness of a preoperative noninvasive lymph node staging (NILS) model (an artificial neural network model) for predicting pathological nodal status in patients with cN0 breast cancer (BC). METHODS: A health-economic decision-analytic model was developed to evaluate the utility of the NILS model in reducing the proportion of cN0 patients with low predicted risk undergoing SLNB. The model used information from a national registry and published studies, and three sensitivity/specificity scenarios of the NILS model were evaluated. Subgroup analysis explored the outcomes of breast-conserving surgery (BCS) or mastectomy. The results are presented as cost (€) and quality-adjusted life years (QALYs) per 1000 patients. RESULTS: All three scenarios of the NILS model reduced total costs (-€93,244 to -€398,941 per 1000 patients). The overall health benefit allowing for the impact of SLNB complications was a net health gain (7.0-26.9 QALYs per 1000 patients). Sensitivity analyses disregarding reduced quality of life from lymphedema showed a small loss in total health benefits (0.4-4.0 QALYs per 1000 patients) because of the reduction in total life years (0.6-6.5 life years per 1000 patients) after reduced adjuvant treatment. Subgroup analyses showed greater cost reductions and QALY gains in patients undergoing BCS. CONCLUSION: Implementing the NILS model to identify patients with low risk for nodal metastases was associated with substantial cost reductions and likely overall health gains, especially in patients undergoing BCS.

The association between body mass index and pathological complete response in neoadjuvant-treated breast cancer patients.

BACKGROUND: Obesity seems to be associated with a poorer response to adjuvant chemotherapy in breast cancer (BC); however, associations in the neoadjuvant chemotherapy (NACT) setting and according to menopausal status are less studied. This study aims to investigate the association between pretreatment body mass index (BMI) and pathological complete response (pCR) following NACT in BC according to menopausal and estrogen receptor (ER) status. MATERIAL AND METHODS: The study cohort consisted of 491 patients receiving NACT in 2005-2019. Based on pre-NACT patient and tumor characteristics, the association between BMI and achieving pCR was analyzed using logistic regression models (crude and adjusted models (age, tumor size, and node status)) with stratification by menopausal and ER status. RESULTS: In the overall cohort, being overweight (BMI ≥25) compared by being normal-weight (BMI <25), increased the odds of accomplishing pCR by 15%. However, based on the 95% confidence interval (CI) the data were compatible with associations within the range of a decrease of 30% to an increase of 89%. Stratification according to menopausal status also showed no strong association: the odds ratio (OR) of accomplishing pCR in overweight premenopausal patients compared with normal-weight premenopausal patients was 1.76 (95% CI 0.88-3.55), whereas for postmenopausal patients the corresponding OR was 0.71 (95% CI 0.35-1.46). DISCUSSION: In a NACT BC cohort of 491 patients, we found no evidence of high BMI as a predictive factor of accomplishing pCR, neither in the whole cohort nor stratified by menopausal status. Given the limited precision in our results, larger studies are needed before considering BMI in clinical decision-making regarding NACT or not.

The NILS Study Protocol: A Retrospective Validation Study of an Artificial Neural Network Based Preoperative Decision-Making Tool for Noninvasive Lymph Node Staging in Women with Primary Breast Cancer (ISRCTN14341750).

Newly diagnosed breast cancer (BC) patients with clinical T1-T2 N0 disease undergo sentinel-lymph-node (SLN) biopsy, although most of them have a benign SLN. The pilot noninvasive lymph node staging (NILS) artificial neural network (ANN) model to predict nodal status was published in 2019, showing the potential to identify patients with a low risk of SLN metastasis. The aim of this study is to assess the performance measures of the model after a web-based implementation for the prediction of a healthy SLN in clinically N0 BC patients. This retrospective study was designed to validate the NILS prediction model for SLN status using preoperatively available clinicopathological and radiological data. The model results in an estimated probability of a healthy SLN for each study participant. Our primary endpoint is to report on the performance of the NILS prediction model to distinguish between healthy and metastatic SLNs (N0 vs. N+) and compare the observed and predicted event rates of benign SLNs. After validation, the prediction model may assist medical professionals and BC patients in shared decision making on omitting SLN biopsies in patients predicted to be node-negative by the NILS model. This study was prospectively registered in the ISRCTN registry (identification number: 14341750).

Predicting pathological axillary lymph node status with ultrasound following neoadjuvant therapy for breast cancer.

PURPOSE: High-performing imaging and predictive markers are warranted to minimize surgical overtreatment of the axilla in breast cancer (BC) patients receiving neoadjuvant chemotherapy (NACT). Here we have investigated whether axillary ultrasound (AUS) could identify axillary lymph node (ALN) metastasis (ALNM) pre-NACT and post-NACT for BC. The association of tumor, AUS features and mammographic density (MD) with axillary-pathological complete response (axillary-pCR) post-NACT was also assessed. METHODS: The NeoDense-study cohort (N = 202, NACT during 2014-2019), constituted a pre-NACT cohort, whereas patients whom had a cytology verified ALNM pre-NACT and an axillary dissection performed (N = 114) defined a post-NACT cohort. AUS characteristics were prospectively collected pre- and post-NACT. The diagnostic accuracy of AUS was evaluated and stratified by histological subtype and body mass index (BMI). Predictors of axillary-pCR were analyzed, including MD, using simple and multivariable logistic regression models. RESULTS: AUS demonstrated superior performance for prediction of ALNM pre-NACT in comparison to post-NACT, as reflected by the positive predictive value (PPV) 0.94 (95% CI 0.89-0.97) and PPV 0.76 (95% CI 0.62-0.87), respectively. We found no difference in AUS performance according to neither BMI nor histological subtype. Independent predictors of axillary-pCR were: premenopausal status, ER-negativity, HER2-overexpression, and high MD. CONCLUSION: Baseline AUS could, to a large extent, identify ALNM; however, post-NACT, AUS was insufficient to determine remaining ALNM. Thus, our results support the surgical staging of the axilla post-NACT. Baseline tumor biomarkers and patient characteristics were predictive of axillary-pCR. Larger, multicenter studies are needed to evaluate the performance of AUS post-NACT.

Mammographic density as an image-based biomarker of therapy response in neoadjuvant-treated breast cancer patients.

PURPOSE: Personalized cancer treatment requires predictive biomarkers, including image-based biomarkers. Breast cancer (BC) patients receiving neoadjuvant chemotherapy (NACT) are in a clinically vulnerable situation with the tumor present. This study investigated whether mammographic density (MD), assessed pre-NACT, is predictive of pathological complete response (pCR). METHODS: A total of 495 BC patients receiving NACT in Sweden 2005-2019 were included, merged from two different cohorts. Cohort 1 was retrospectively collected (n = 295) and cohort 2 was prospectively collected (n = 200). Mammograms were scored for MD pre-NACT according to the Breast Imaging-Reporting and Data System (BI-RADS), 5th Edition. The association between MD and accomplishing pCR post-NACT was analyzed using logistic regression models-for the whole cohort, stratified by menopausal status, and in different St. Gallen surrogate subtypes. RESULTS: In comparison to patients with low MD (BI-RADS a), the multivariable-adjusted odds ratio (OR) of accomplishing pCR following NACT was on a descending scale: 0.62 (95% confidence interval (CI) 0.24-1.57), 0.38 (95% CI 0.14-1.02), and 0.32 (95% CI 0.09-1.08) for BI-RADS b, c, and d, respectively. For premenopausal patients selectively, the corresponding point estimates were lower, although wider CIs: 0.31 (95% CI 0.06-1.62), 0.24 (95% CI 0.04-1.27), and 0.13 (95% CI 0.02-0.88). Subgroup analyses based on BC subtypes resulted in imprecise estimates, i.e., wide CIs. CONCLUSIONS: It seemed as though patients with higher MD at baseline were less likely to reach pCR after NACT-a finding more pronounced in premenopausal women. Larger multicenter studies are needed to enable analyses and interpretation for different BC subtypes.

Neoadjuvant breast cancer treatment response; tumor size evaluation through different conventional imaging modalities in the NeoDense study.

BACKGROUND: Neoadjuvant chemotherapy (NACT) is offered to an increasing number of breast cancer (BC) patients, and comprehensive monitoring of treatment response is of utmost importance. Several imaging modalities are available to follow tumor response, although likely to provide different clinical information. We aimed to examine the association between early radiological response by three conventional imaging modalities and pathological complete response (pCR). Further, we investigated the agreement between these modalities pre-, during, and post-NACT, and the accuracy of predicting pathological residual tumor burden by these imaging modalities post-NACT. MATERIAL AND METHODS: This prospective Swedish cohort study included 202 BC patients assigned to NACT (2014-2019). Breast imaging with clinically used modalities: mammography, ultrasound, and tomosynthesis was performed pre-, during, and post-NACT. We investigated the agreement of tumor size by the different imaging modalities, and their accuracy of tumor size estimation. Patients with a radiological complete response or radiological partial response (≥30% decrease in tumor diameter) during NACT were classified as radiological early responders. RESULTS: Patients with an early radiological response by ultrasound had 2.9 times higher chance of pCR than early radiological non-responders; the corresponding relative chance for mammography and tomosynthesis tumor size measures was 1.8 and 2.8, respectively. Post-NACT, each modality, separately, could accurately estimate tumor size (within 5 mm margin compared to pathological evaluation) in 43-46% of all tumors. The diagnostic precision in predicting pCR post-NACT was similar between the three imaging modalities; however, tomosynthesis had slightly higher specificity and positive predictive values. CONCLUSION: Breast imaging modalities correctly estimated pathological tumor size in less than half of the tumors. Based on this finding, predicting residual tumor size post-NACT is challenging using conventional imaging. Patients with early radiological non-response might need improved monitoring during NACT and be considered for changed treatment plans.

Mammographic density changes during neoadjuvant breast cancer treatment: NeoDense, a prospective study in Sweden.

OBJECTIVES: To assess if mammographic density (MD) changes during neoadjuvant breast cancer treatment and is predictive of a pathological complete response (pCR). METHODS: We prospectively included 200 breast cancer patients assigned to neoadjuvant chemotherapy (NACT) in the NeoDense study (2014-2019). Raw data mammograms were used to assess MD with a fully automated volumetric method and radiologists categorized MD using the Breast Imaging-Reporting and Data System (BI-RADS), 5th Edition. Logistic regression was used to calculate odds ratios (OR) for pCR comparing BI-RADS categories c vs. a, b, and d as well as with a 0.5% change in percent dense volume adjusting for baseline characteristics. RESULTS: The overall median age was 53.1 years, and 48% of study participants were premenopausal pre-NACT. A total of 23% (N = 45) of the patients accomplished pCR following NACT. Patients with very dense breasts (BI-RADS d) were more likely to have a positive axillary lymph node status at diagnosis: 89% of the patients with very dense breasts compared to 72% in the entire cohort. A total of 74% of patients decreased their absolute dense volume during NACT. The likelihood of accomplishing pCR following NACT was independent of volumetric MD at diagnosis and change in volumetric MD during treatment. No trend was observed between decreasing density according to BI-RADS and the likelihood of accomplishing pCR following NACT. CONCLUSIONS: The majority of patients decreased their MD during NACT. We found no evidence of MD as a predictive marker of pCR in the neoadjuvant setting.

Mammographic density is a potential predictive marker of pathological response after neoadjuvant chemotherapy in breast cancer.

BACKGROUND: Our aim is to study if mammographic density (MD) prior to neoadjuvant chemotherapy is a predictive factor in accomplishing a pathological complete response (pCR) in neoadjuvant-treated breast cancer patients. METHODS: Data on all neoadjuvant treated breast cancer patients in Southern Sweden (2005-2016) were retrospectively identified, with patient and tumor characteristics retrieved from their medical charts. Diagnostic mammograms were used to evaluate and score MD as categorized by breast composition with the Breast Imaging-Reporting and Data System (BI-RADS) 5th edition. Logistic regression was used in complete cases to assess the odds ratios (OR) for pCR compared to BI-RADS categories (a vs b-d), adjusting for patient and pre-treatment tumor characteristics. RESULTS: A total of 302 patients were included in the study population, of which 57 (18.9%) patients accomplished pCR following neoadjuvant chemotherapy. The number of patients in the BI-RADS category a, b, c, and d were separately 16, 120, 140, and 26, respectively. In comparison to patients with BI-RADS breast composition a, patients with denser breasts had a lower OR of accomplishing pCR: BI-RADS b 0.32 (95%CI 0.07-0.1.5), BI-RADS c 0.30 (95%CI 0.06-1.45), and BI-RADS d 0.06 (95%CI 0.01-0.56). These associations were measured with lower point estimates, but wider confidence interval, in premenopausal patients; OR of accomplishing pCR for BI-RADS d in comparison to BI-RADS a: 0.03 (95%CI 0.00-0.76). CONCLUSIONS: The likelihood of accomplishing pCR is indicated to be lower in breast cancer patients with higher MD, which need to be analysed in future studies for improved clinical decision-making regarding neoadjuvant treatment.

Association between uterine artery Doppler blood flow changes and arterial wall elasticity in pregnant women.

INTRODUCTION: Uterine artery (UtA) Doppler velocimetry changes and increased arterial stiffness are associated with preeclampsia. We aimed to investigate the relation between UtA velocimetry changes and arterial stiffness in pregnant women. METHODS: Doppler velocimetry and photoplethysmographic digital pulse wave analysis (DPA) were performed in 173 pregnant women in the second or the third trimester, where UtA Doppler pulsatility index (PI), diastolic notching, and UtA score (UAS) combining notching and high PI were calculated. DPA stiffness parameters representing large arteries were ejection elasticity index (EEI) and b/a, small arteries dicrotic index (DI) and d/a, and global stiffness the aging index (AI). RESULTS: One hundred and thirty women had normal Doppler and 43 had diastolic notching, of whom nine had high PI. DI indicated increased stiffness in small arteries when notching was present (p = 0.044) and showed a significant but weak correlation to UAS (p = 0.025, tau 0.12). EEI and b/a indicated increased large artery stiffness (p ≤0.014), d/a small artery stiffness (p = 0.023), and AI a systemic stiffness (p = 0.040) when high PI. CONCLUSION: High UtA PI was associated with increased systemic arterial stiffness, whereas notching was related to increased stiffness in small arteries only. This indicates pathophysiological differences between the two Doppler parameters.

Effects of statin use on volumetric mammographic density: results from the KARMA study.

BACKGROUND: Epidemiological data on statins and breast cancer risk have been inconclusive. The aim of this study was to clarify the role of statins in breast cancer risk by studying their effect on mammographic density. METHODS: The KARolinska MAmmography project for risk prediction of breast cancer (KARMA) includes 70,877 women who underwent either a screening or clinical mammography from January 2011 to December 2013. In total, 41,102 women responded to a web-based questionnaire, and had raw digital mammograms stored. Volumetric mammographic density was measured using Volpara™ and information on statin use was obtained through linkage with the Swedish National Prescription Register. Analysis of covariance was used to study the effect of statin use on mammographic density, adjusting for a large set of potential confounders. We also studied the effects of statin class and treatment duration and tested for potential effect modification by hormone replacement therapy (HRT). RESULTS: Statin use was recorded in 3,337 women (8.1 %) of the study population and lipophilic statins was the most commonly prescribed type (93.4 % of all statin users). After multivariable adjustment, percent dense volume was lower in statin users than in non-users (P < 0.001). This association was explained by a larger absolute non-dense volume in statin users (P < 0.001). Overall, no difference in absolute dense volume was detected, but interaction analyses revealed a larger dense volume among statin users who reported ever HRT use (P = 0.03). No differential effects were observed according to statin lipophilicity and treatment duration. CONCLUSIONS: We observed no overall effect of statin use on mammographic density in terms of absolute dense volume, although a larger absolute dense volume was observed in statin users who reported ever HRT use, which requires further investigation.