The region-of-interest size impacts on Ki67 quantification by computer-assisted image analysis in breast cancer

Applications of Artificial Intelligence in Breast Pathology

CONTEXT.—

Increasing implementation of whole slide imaging together with digital workflow and advances in computing capacity enable the use of artificial intelligence (AI) in pathology, including breast pathology. Breast pathologists often face a significant workload, with diagnosis complexity, tedious repetitive tasks, and semiquantitative evaluation of biomarkers. Recent advances in developing AI algorithms have provided promising approaches to meet the demand in breast pathology.

OBJECTIVE.—

To provide an updated review of AI in breast pathology. We examined the success and challenges of current and potential AI applications in diagnosing and grading breast carcinomas and other pathologic changes, detecting lymph node metastasis, quantifying breast cancer biomarkers, predicting prognosis and therapy response, and predicting potential molecular changes.

DATA SOURCES.—

We obtained data and information by searching and reviewing literature on AI in breast pathology from PubMed and based our own experience.

CONCLUSIONS.—

With the increasing application in breast pathology, AI not only assists in pathology diagnosis to improve accuracy and reduce pathologists' workload, but also provides new information in predicting prognosis and therapy response.

Noninferiority of Artificial Intelligence-Assisted Analysis of Ki-67 and Estrogen/Progesterone Receptor in Breast Cancer Routine Diagnostics

Image analysis assistance with artificial intelligence (AI) has become one of the great promises over recent years in pathology, with many scientific studies being published each year. Nonetheless, and perhaps surprisingly, only few image AI systems are already in routine clinical use. A major reason for this is the missing validation of the robustness of many AI systems: beyond a narrow context, the large variability in digital images due to differences in preanalytical laboratory procedures, staining procedures, and scanners can be challenging for the subsequent image analysis. Resulting faulty AI analysis may bias the pathologist and contribute to incorrect diagnoses and, therefore, may lead to inappropriate therapy or prognosis. In this study, a pretrained AI assistance tool for the quantification of Ki-67, estrogen receptor (ER), and progesterone receptor (PR) in breast cancer was evaluated within a realistic study set representative of clinical routine on a total of 204 slides (72 Ki-67, 66 ER, and 66 PR slides). This represents the cohort with the largest image variance for AI tool evaluation to date, including 3 staining systems, 5 whole-slide scanners, and 1 microscope camera. These routine cases were collected without manual preselection and analyzed by 10 participant pathologists from 8 sites. Agreement rates for individual pathologists were found to be 87.6% for Ki-67 and 89.4% for ER/PR, respectively, between scoring with and without the assistance of the AI tool regarding clinical categories. Individual AI analysis results were confirmed by the majority of pathologists in 95.8% of Ki-67 cases and 93.2% of ER/PR cases. The statistical analysis provides evidence for high interobserver variance between pathologists (Krippendorff's α, 0.69) in conventional immunohistochemical quantification. Pathologist agreement increased slightly when using AI support (Krippendorff α, 0.72). Agreement rates of pathologist scores with and without AI assistance provide evidence for the reliability of immunohistochemical scoring with the support of the investigated AI tool under a large number of environmental variables that influence the quality of the diagnosed tissue images.

p53 Expression in Luminal Breast Cancer Correlates With TP53 Mutation and Primary Endocrine Resistance

TP53 mutation is associated with primary endocrine resistance in luminal breast cancer (BC). Nuclear accumulation of p53, as determined by immunohistochemistry (IHC), is a surrogate marker for TP53 mutation. The immunohistochemical p53 index that defines a p53-positive status is not well established. This study determined the optimal p53 index cutoff to identify luminal BCs harboring TP53 mutations. In total, 364 luminal BCs from the West German Study Group ADAPT trial (NCT01779206) were analyzed for TP53 mutations by next-generation sequencing and for p53 expression by IHC (DO-7 antibody). P53 indices were determined by automated image analysis. All tumors were from patients treated with short-term preoperative endocrine therapy (pET; tamoxifen or aromatase inhibitor) before tumor resection. IHC evaluation included needle biopsies before therapy (baseline) and resections specimens after therapy (post-pET). Optimal p53 index cutoffs were defined with Youden statistics. TP53 mutations were detected in 16.3% of BC cases. The median p53 indices were significantly higher in TP53-mutated BCs compared to BCs harboring wild-type TP53 (baseline: 47.0% vs 6.4%, P < .001; post-pET: 50.1% vs 1.1%, P < .001). Short-term pET decreased p53 indices in BCs harboring wild-type TP53 (P < .001) but not in TP53-mutated BCs (P = .102). For baseline biopsies, the optimal p53 index cutoff was ≥34.6% (specificity 0.92, sensitivity 0.63, Youden index 0.54, accuracy: 0.87). For post-pET specimens, the optimal cutoff was ≥25.3% (specificity 0.95, sensitivity 0.65, Youden index 0.60, accuracy: 0.90). Using these cutoffs to define the p53 status, p53-positive BCs were >2-fold more common in pET nonresponders compared to pET responders (baseline: 37/162, 22.8% vs 18/162, 11.1%, P = .007; post-pET: 36/179, 20.1% vs 16/179, 8.9%, P = .004). In summary, IHC for p53 identifies TP53-mutated luminal BCs with high specificity and accuracy. Optimal cutoffs are ≥35% and ≥25% for treatment-naïve and endocrine-pretreated patients, respectively.

Clinical validity and clinical utility of Ki67 in early breast cancer

Ki67 represents an immunohistochemical nuclear localized marker that is widely used in surgical pathology. Nuclear immunoreactivity for Ki67 indicates that cells are cycling and are in G1- to S-phase. The percentage of Ki67-positive tumor cells (Ki67 index) therefore provides an estimate of the growth fraction in tumor specimens. In breast cancer (BC), tumor cell proliferation rate is one of the most relevant prognostic markers and Ki67 is consequently helpful in prognostication similar to histological grading and mRNA profiling-based BC risk stratification. In BCs treated with short-term preoperative endocrine therapy, Ki67 dynamics enable distinguishing between endocrine sensitive and resistant tumors. Despite its nearly universal use in pathology laboratories worldwide, no internationally accepted consensus has yet been achieved for some methodological details related to Ki67 immunohistochemistry (IHC). Controversial issues refer to choice of IHC antibody clones, scoring methods, inter-laboratory reproducibility, and the potential value of computer-assisted imaging analysis and/or artificial intelligence for Ki67 assessment. Prospective clinical trials focusing on BC treatment have proven that Ki67, as determined by standardized central pathology assessment, is of clinical validity. Clinical utility has been demonstrated in huge observational studies.

Endocrine Therapy Response and 21-Gene Expression Assay for Therapy Guidance in HR+/HER2- Early Breast Cancer

PURPOSE

To our knowledge, WSG-ADAPT-HR+/HER2- (NCT01779206; n = 5,625 registered) is the first trial combining the 21-gene expression assay (recurrence score [RS]) and response to 3-week preoperative endocrine therapy (ET) to guide systemic therapy in early breast cancer.

MATERIALS AND METHODS

Baseline and postendocrine Ki67 (Ki67_post) were evaluated centrally. In the endocrine trial, all patients received exclusively ET: patients with pathologic regional lymph node status (pN) 0-1 (ie, 0-3 involved lymph nodes) entered control arm if RS ≤ 11 and experimental arm if RS12-25 with ET response (Ki67_post ≤ 10%). All other patients (including N0-1 RS12-25 without ET response) received dose-dense chemotherapy (CT) followed by ET in the CT trial. Primary end point of the endocrine trial was noninferiority of 5-year invasive disease-free survival (5y-iDFS) in experimental (v control) arm; secondary end points included distant DFS, overall survival, and translational research.

RESULTS

Intention-to-treat population comprised 2,290 patients (n = 1,422 experimental v n = 868 control): 26.3% versus 34.6% premenopausal and 27.4% versus 24.0% pN1. One-sided 95% lower confidence limit of the 5y-iDFS difference was -3.3%, establishing prespecified noninferiority (P = .05). 5y-iDFS was 92.6% (95% CI, 90.8 to 94.0) in experimental versus 93.9% (95% CI, 91.8 to 95.4) in control arm; 5-year distant DFS was 95.6% versus 96.3%, and 5-year overall survival 97.3% versus 98.0%, respectively. Differences were similar in age and nodal subgroups. In N0-1 RS12-25, outcome of ET responders (ET alone) was comparable with that of ET nonresponders (CT) for age > 50 years and superior for age ≤ 50 years. ET response was more likely with aromatase inhibitors (mostly postmenopausal) than with tamoxifen (mostly premenopausal): 78.1% versus 41.1% (P < .001). ET response was 78.8% in RS0-11, 62.2% in RS12-25, and 32.7% in RS > 25 (n = 4,203, P < .001).

CONCLUSION

WSG-ADAPT-HR+/HER2- demonstrates that guiding systemic treatment by both RS and ET response is feasible in clinical routine and spares CT in pre- and postmenopausal patients with ≤ 3 involved lymph nodes.

ERBB2 mutation is associated with sustained tumor cell proliferation after short-term preoperative endocrine therapy in early lobular breast cancer

Invasive lobular breast cancer (ILC) is a special breast cancer (BC) subtype and is mostly hormone receptor (HR)-positive and ERBB2 non-amplified. Endocrine therapy restrains tumor proliferation and is the mainstay of lobular BC treatment. Mutation of ERBB2 has been associated with recurrent ILC. However, it is unknown whether ERBB2 mutation impacts on the otherwise exquisite responsiveness of early ILC to endocrine therapy. We have recently profiled n = 622 HR-positive early BCs from the ADAPT trial for mutations in candidate genes involved in endocrine resistance, including ERBB2. All patients were treated with short-term preoperative endocrine therapy (pET, tamoxifen or aromatase inhibitors) before tumor resection. Tumor proliferation after endocrine therapy (post-pET Ki67 index) was determined prospectively by standardized central pathology assessment supported by computer-assisted image analysis. Sustained or suppressed proliferation were defined as post-pET Ki67 ≥10% or <10%. Here, we report a subgroup analysis pertaining to ILCs in this cohort. ILCs accounted for 179/622 (28.8%) cases. ILCs were enriched in mutations in CDH1 (124/179, 69.3%, P < 0.0001) and ERBB2 (14/179, 7.8%, P < 0.0001), but showed fewer mutations in TP53 (7/179, 3.9%, P = 0.0048) and GATA3 (11/179, 6.1%, P < 0.0001). Considering all BCs irrespective of subtypes, ERBB2 mutation was not associated with proliferation. In ILCs, however, ERBB2 mutations were 3.5-fold more common in cases with sustained post-pET proliferation compared to cases with suppressed post-pET proliferation (10/75, 13.3% versus 4/104, 3.8%, P = 0.0248). Moreover, ERBB2 mutation was associated with high Oncotype DX recurrence scores (P = 0.0087). In summary, our findings support that ERBB2 mutation influences endocrine responsiveness in early lobular BC.

TP53 mutations are associated with primary endocrine resistance in luminal early breast cancer

Background

Whereas the genomic landscape of endocrine‐resistant breast cancer has been intensely characterized in previously treated cases with local or distant recurrence, comparably little is known about genomic alterations conveying primary non‐responsiveness to endocrine treatment in luminal early breast cancer.

Methods

In this study, 622 estrogen receptor‐expressing breast cancer cases treated with short‐term preoperative endocrine therapy (pET) from the WSG‐ADAPT trial (NCT01779206) were analyzed for genetic alterations associated with impaired endocrine proliferative response (EPR) to 3‐week pET with tamoxifen or aromatase inhibitors. EPR was categorized as optimal (post‐pET Ki67 <10%) versus slightly, moderately, and severely impaired (post‐pET Ki67 10%–19%, 20%–34%, and ≥35%, respectively). Recently described gene mutations frequently found in previously treated advanced breast cancer were analyzed (ARID1A, BRAF, ERBB2, ESR1, GATA3, HRAS, KRAS, NRAS, PIK3CA, and TP53) by next‐generation sequencing. Amplifications of CCND1, FGFR1, ERBB2, and PAK1 were determined by digital PCR or fluorescence in situ hybridization.

Results

ERBB2 amplification (p = 0.0015) and mutations of TP53 (p < 0.0001) were significantly associated with impaired EPR. Impaired EPR in TP53‐mutated breast cancer cases was independent from the Oncotype DX Recurrence Score group and was seen both with tamoxifen‐ and aromatase inhibitor‐based pET (p = 0.0005 each).

Conclusion

We conclude that impaired EPR to pET is suitable to identify cases with primary endocrine resistance in early luminal breast cancer and that TP53‐mutated luminal cancers might not be sufficiently treated by endocrine therapy alone.

Machine-Learning-Based Evaluation of Intratumoral Heterogeneity and Tumor-Stroma Interface for Clinical Guidance

Assessment of intratumoral heterogeneity and tumor-host interaction within the tumor microenvironment is becoming increasingly important for innovative cancer therapy decisions because of the unique information it can generate about the state of the disease. However, its assessment and quantification are limited by ambiguous definitions of the tumor-host interface and by human cognitive capacity in current pathology practice. Advances in machine learning and artificial intelligence have opened the field of digital pathology to novel tissue image analytics and feature extraction for generation of high-capacity computational disease management models. A particular benefit is expected from machine-learning applications that can perform extraction and quantification of subvisual features of both intratumoral heterogeneity and tumor microenvironment aspects. These methods generate information about cancer cell subpopulation heterogeneity, potential tumor-host interactions, and tissue microarchitecture, derived from morphologically resolved content using both explicit and implicit features. Several studies have achieved promising diagnostic, prognostic, and predictive artificial intelligence models that often outperform current clinical and pathology criteria. However, further effort is needed for clinical adoption of such methods through development of standardizable high-capacity workflows and proper validation studies.

Improving Ki67 assessment concordance by the use of an artificial intelligence-empowered microscope: a multi-institutional ring study

AIMS

The nuclear proliferation biomarker Ki67 plays potential prognostic and predictive roles in breast cancer treatment. However, the lack of interpathologist consistency in Ki67 assessment limits the clinical use of Ki67. The aim of this article was to report a solution utilising an artificial intelligence (AI)-empowered microscope to improve Ki67 scoring concordance.

METHODS AND RESULTS

We developed an AI-empowered microscope in which the conventional microscope was equipped with AI algorithms, and AI results were provided to pathologists in real time through augmented reality. We recruited 30 pathologists with various experience levels from five institutes to assess the Ki67 labelling index on 100 Ki67-stained slides from invasive breast cancer patients. In the first round, pathologists conducted visual assessment on a conventional microscope; in the second round, they were assisted with reference cards; and in the third round, they were assisted with an AI-empowered microscope. Experienced pathologists had better reproducibility and accuracy [intraclass correlation coefficient (ICC) = 0.864, mean error = 8.25%] than inexperienced pathologists (ICC = 0.807, mean error = 11.0%) in visual assessment. Moreover, with reference cards, inexperienced pathologists (ICC = 0.836, mean error = 10.7%) and experienced pathologists (ICC = 0.875, mean error = 7.56%) improved their reproducibility and accuracy. Finally, both experienced pathologists (ICC = 0.937, mean error = 4.36%) and inexperienced pathologists (ICC = 0.923, mean error = 4.71%) improved the reproducibility and accuracy significantly with the AI-empowered microscope.

CONCLUSION

The AI-empowered microscope allows seamless integration of the AI solution into the clinical workflow, and helps pathologists to obtain higher consistency and accuracy for Ki67 assessment.

Spatial Distribution and Prognostic Implications of Tumor-Infiltrating FoxP3- CD4+ T Cells in Biliary Tract Cancer

PURPOSE

The clinical implications of tumor-infiltrating T cell subsets and their spatial distribution in biliary tract cancer (BTC) patients treated with gemcitabine plus cisplatin were investigated.

MATERIALS AND METHODS

A total of 52 BTC patients treated with palliative gemcitabine plus cisplatin were included. Multiplexed immunohistochemistry was performed on tumor tissues, and immune infiltrates were separately analyzed for the stroma, tumor margin, and tumor core.

RESULTS

The density of CD8+ T cells, FoxP3- CD4+ helper T cells, and FoxP3+ CD4+ regulatory T cells was significantly higher in the tumor margin than in the stroma and tumor core. The density of LAG3- or TIM3-expressing CD8+ T cell and FoxP3- CD4+ helper T cell infiltrates was also higher in the tumor margin. In extrahepatic cholangiocarcinoma, there was a higher density of T cell subsets in the tumor core and regulatory T cells in all regions. A high density of FoxP3- CD4+ helper T cells in the tumor margin showed a trend toward better progression-free survival (PFS) (p=0.092) and significantly better overall survival (OS) (p=0.012). In multivariate analyses, a high density of FoxP3- CD4+ helper T cells in the tumor margin was independently associated with favorable PFS and OS.

CONCLUSION

The tumor margin is the major site for the active infiltration of T cell subsets with higher levels of LAG3 and TIM3 expression in BTC. The density of tumor margin-infiltrating FoxP3- CD4+ helper T cells may be associated with clinical outcomes in BTC patients treated with gemcitabine plus cisplatin.

Ki67 index and mitotic count: Correlation and variables affecting the accuracy of the quantification in endocrine/neuroendocrine tumors

Quantification of Ki67 and mitosis is time consuming and subject to inter-observer variabilities. Limited studies explored the impact of those variables on the results and the correlation between mitotic count and Ki67 index in endocrine/neuroendocrine tumors, particularly so since the advent of PHH3 antibody and digital pathology. Using Ki67 and mitosis as examples, this study is intended to reveal variables affecting accurate quantification of biomarkers, and to explore the relationship of Ki67 index and mitotic count/index in endocrine/neuroendocrine tumors. Using both manual and pathologist supervised digital image analysis (PSDIA) methods, we examined the impact of post-analytical variables on the quantification of mitosis and Ki67 index and studied the correlation between them in 41 cases of endocrine/neuroendocrine tumors of variable histological grades/proliferating rates. We found that the selection of hotspots, field size and especially threshold affected the outcome of quantification of mitosis and Ki67 index; that mitotic count/index strongly (p < 0.05) correlated with Ki67 index only in the tumors with peak Ki67 index less than 30% and the correlation was more monotonic (positive, non-linear) than linear. In the hotspots of these tumors, the ratio of mitotic count to proliferating cells defined by Ki67 detection averaged 0.04. We also found that the PHH3 antibody could markedly increase the efficiency and accuracy of mitotic quantification. A consensus among pathologists is needed for the selection of hotspots, field size and threshold for quantification of mitosis and Ki67 index.

Sequential immunohistochemistry and virtual image reconstruction using a single slide for quantitative KI67 measurement in breast cancer

Objective

Ki67 is a prognostic and predictive marker in breast cancer (BC). However, manual scoring (MS) by visual assessment suffers from high inter-observer variability which limits its clinical use. Here, we developed a new digital image analysis (DIA) workflow, named KiQuant for automated scoring of Ki67 and investigated its equivalence with standard pathologist's assessment.

Methods

Sequential immunohistochemistry of Ki67 and cytokeratin, for precise tumor cell recognition, were performed in the same section of 5 tissue microarrays containing 329 tumor cores from different breast cancer subtypes. Slides were digitalized and subjected to DIA and MS for Ki67 assessment. The intraclass correlation coefficient (ICC) and Bland-Altman plot were used to evaluate inter-observer reproducibility. The Kaplan-Meier analysis was used to determine the prognostic potential.

Results

KiQuant showed an excellent correlation with MS (ICC:0.905,95%CI:0.878–0.926) with satisfactory inter-run (ICC:0.917,95%CI:0.884–0.942) and inter-antibody reproducibilities (ICC:0.886,95%CI:0.820–0.929). The distance between KiQuant and MS increased with the magnitude of Ki67 measurement and positively correlated with analyzed tumor area and breast cancer subtype. Agreement rates between KiQuant and MS within the clinically relevant 14% and 30% cut-off points ranged from 33% to 44% with modest interobserver reproducibility below the 20% cut-off (0.606, 95%CI:0.467–0.727). High Ki67 by KiQuant correlated with worse outcome in all BC and in the luminal subtype (P = 0.028 and P = 0.043, respectively). For MS, the association with survival was significant only in 1 out of 3 observers.

Conclusions

KiQuant represents an easy and accurate methodology for Ki67 measurement providing a step toward utilizing Ki67 in the clinical setting.

•

Automated Ki67 scoring workflow improved reproducibility.

•

Sequential immunohistochemistry in the same section for precise cell recognition.

•

Use of a tumor mask for automatic tumor region selection.

•

Outperform pathologist-based Ki67 scoring in prognostic prediction.

Reproducibility of histologic prognostic parameters for mantle cell lymphoma: cytology, Ki67, p53 and SOX11

Mantle cell lymphoma (MCL) shows a clinical aggressiveness that varies from patient to patient. Despite major advances in outcomes with current immunochemotherapy, the future development of therapies requires risk stratification to tailor therapy intensity. Within the group of reference pathologists for the ongoing trials of the European MCL Network, we performed a round robin test on a tissue microarray to evaluate the reproducibility in assessing the biomarkers of outcome in MCL. Cytological subtype, Ki67-index and expression of p53 and SOX11 were evaluated on 20 diagnostic tumour samples by eight participating labs independently. We demonstrate that the assessment of the proliferation index by counting the Ki67 positive cells as well as assessment of SOX11 and p53 expression status is reproducible between labs. For the most established prognostic biomarker, Ki67, the intra-class correlation coefficient was very good when assessed as a continuous parameter (0.87). The agreement was lower when the values were analysed in a dichotomized way applying the commonly used cutoff of 30% (kappa = 0.65, complete concordance of all labs in 13/20 (65%)). Cases with discrepant results between labs in the dichotomized analysis showed mean values close to the cutoff of 30%. Centralised scoring and digital image analysis revealed results in line with the scores from individual labs. All cases in our cohort were additionally assessed for gene expression signatures and of TP53 gene alterations. Given the good reproducibility when guidelines of assessment are applied, the biomarker studied in this inter-laboratory test presents potential candidates to be enhanced for risk-stratification in the future clinical trials.

An interobserver reproducibility analysis of size-set semiautomatic counting for Ki67 assessment in breast cancer

Purpose

The proliferation marker Ki67 has prognostic and predictive values in breast cancer, and the cutoff of the Ki67 label index (LI) is a key index for chemotherapy. However, poor interobserver consistency in Ki67 assessment has limited the clinical use of Ki67, especially in luminal cancers. Here, we reported a modified Ki67 assessment method, size-set semiautomatic counting (SSSAC) and investigated its interobserver reproducibility.

Methods

One hundred invasive breast cancer tissues were set immunostained for Ki67 in one laboratory, scanned as digital slides, and sent to 41 pathologists at the laboratories of 16 hospitals for Ki67 LI assessment using size-set semiautomatic counting (SSSAC), size-set visual assessment (SSVA) and size-set digital image analysis (SSDIA) with a specific image viewing software (Aperio Image Scope, Leica, Germany). The intraclass correlation coefficient (ICC) and Bland-Altman plot were used to evaluate interobserver reproducibility. The Wilcoxon signed-rank test was used to analyze the difference in the Ki67 values assessed by SSSAC and SSDIA.

Results

SSSAC demonstrated better interobserver reproducibility (ICC = 0.942) than SSVA (ICC = 0.802). The interobserver reproducibility was better in Ki67 homogeneously stained slides and centralized hot-spot slides than in scattered hot-spot slides. The Ki67 value assessed with SSSAC was obviously higher than that assessed with SSDIA (negative ranks (SSDIA < SSSAC): N = 80, sum of ranks = 4274.50; positive ranks (SSDIA > SSSAC): N = 17, sum of ranks = 478.50; Z = −6.837; P < 0.001).

Conclusion

SSSAC shows satisfactory interobserver reproducibility in the Ki67 assessment of breast cancer and may be a candidate standard method for Ki67 LI assessment in breast cancer and other malignancies.

•

A modified method for Ki67 LI assessment based on digital image, size-set semiautomatic counting (SSSAC).

•

Investigated the interobserver reproducibility of SSSAC for Ki67 LI assessment of 100 cases in 41 pathologists of 16 hospitals.

•

SSSAC has better interobserver reproducibility than current visual assessment and was more accurate than DIA in Ki67 assessment.

•

SSSAC is a better choice for Ki67 LI assessment.

Good staining quality ensuring the reproducibility of Ki67 assessment

AIMS

Although Ki67 labelling index (LI) is a prognostic and predictive marker in breast cancer, its accuracy and reproducibility must be validated before its clinical application. We aimed to evaluate the agreement of Ki67 LI in clinical practice in Taiwan.

METHODS

We conducted a Ki67 immunohistochemistry (IHC) proficiency test. The participants performed the Ki67 IHC test and measured the Ki67 LI of 10 cases of breast cancer tissue on a microarray slide. The staining quality was centrally reviewed based on the Ki67 staining of the tonsil surface epithelium.

RESULTS

Ki67 staining and counting methods are diverse in Taiwan. The reproducibility of Ki67 LI was poor to good (intraclass correlation coefficient: 0.581, 95% CI 0.354 to 0.802). The reproducibility and agreement in the high staining quality group were significantly higher than those in the low staining quality group. The majority of the Ki67 LIs derived from the low staining quality group were underestimated. Different counting methods did not reveal significant differences when determining Ki67 LI with microarray sections.

CONCLUSIONS

We suggest using the surface epithelium of the tonsil as external control and achieving optimal staining results that consist of a high positive parabasal layer, a low positive intermediate layer and a negative superficial layer. Good Ki67 staining quality can minimise the staining variations among different laboratories, and it is essential for the reproducibility of Ki67 LI.

Ki-67 assessment in early breast cancer: SAKK28/12 validation study on the IBCSG VIII and IBCSG IX cohort

The assessment of Ki-67 in early-stage breast cancer has become an important diagnostic tool in planning adjuvant therapy, particularly for the administration of additional chemotherapy to hormone-responsive patients. An accurate determination of the Ki-67 index is of the utmost importance; however, the reproducibility is currently unsatisfactory. In this study, we addressed the predictive/prognostic value of Ki-67 index assessed by using the most reproducible methods, which were identified in the pilot phase. Paraffin blocks obtained from patients with moderately differentiated, estrogen receptor (ER)-positive early-stage breast cancer in Switzerland, who were originally randomized to the treatment arms with and without chemotherapy in the IBCSG VIII-IX trials, were retrieved. Of these 344 randomized patients, we identified 158 patients (82 treated with and 76 treated without chemotherapy) for whom sufficient tumour tissue was available. The presence of Ki-67 was assessed visually by counting 2000 cells at the periphery (A) and estimating the number of positive cells in five different peripheral regions (C), which was determined to be the most reproducible method identified the pilot phase. The prognostic and predictive value was assessed by calculating the breast cancer-free interval (BCFI) and overall survival (OS) rate. Ki-67 was considered a numerical and categorical variable when different cut-off values were used (10%, 14%, 20% and 30%). An mRNA-based subtyping by using the MammaTyper kit with the application of a 20% Ki-67 immunohistochemistry (IHC) cut-off equivalent was also performed. 158 of 344 randomized patients could be included in the Ki-67 analysis. The mean Ki-67 values obtained by using the two methods differed (A: 21.32% and C: 16.07%). Ki-67 assessed by using method A with a cut-off of 10% was a predictive marker for OS, as the hazard ratio (>10% vs. <=10%) in patients with chemotherapy was 0.48 with a 95% confidence interval of [0.19–1.19]. Further, the HR of patients treated without chemotherapy was 3.72 with a 95% confidence interval of [1.16–11.96] (p_interaction=0.007). Higher Ki-67 index was not associated with outcome and using the 10% Ki-67 cut-off there was an opposite association for patients with and without chemotherapy. Ki-67 assessments with IHC significantly correlated with MammaTyper results (p=0.002). The exact counting method (A) performed via a light-microscope revealed the predictive value of Ki-67 assessment with a 10% cut-off value. Further analyses employing image analyses and/or mRNA-based-assessments in larger populations are warranted.

[Ki67: biological intertumor variance versus variance of assay]

Since its first description at the Institute of Pathology in Kiel more than 34 years ago, the immunohistochemical proliferation marker Ki67 has been shown to be of prognostic significance in a huge number of retrospective and even some prospective trials on malignant tumours of various tissue derivation. Lack of standardization in the evaluation provides potential sources of variance in assessment. Tumour area to be assessed, minimum number of cells to be analyzed, tedious counting cell by cell or semiquantitative eyeballing, choice of immunohistochemical techniques represent nonstandardized issues that potentially lead to considerable assay heterogeneity. In addition, interpretation is not homogeneous, in particular with regard to thresholds between high and low proliferative activity. Due to these numerous potential methodological limitations, for a long time Ki67 was not generally accepted as a prognostic marker, in particular outside Germany and by nonpathologists. However, in recent years a shift has taken place. Despite the challenge that biological heterogeneity may be hidden by differences in assay performance, Ki67 now plays an important auxiliary role in grading of malignant neoplasms such as breast cancer, neuroendocrine tumours and malignant lymphomas. In this context it is applied in clinical diagnostics as well as in clinical trials for the purpose of stratification. Because of its widespread use, it is of utmost importance to raise awareness of the potential methodological limitations in order to use Ki67 in a meaningful way.

Multiplexed Immunohistochemistry for Molecular and Immune Profiling in Lung Cancer-Just About Ready for Prime-Time?

As targeted molecular therapies and immuno-oncology have become pivotal in the management of patients with lung cancer, the essential requirement for high throughput analyses and clinical validation of biomarkers has become even more intense, with response rates maintained in the 20%–30% range. Moreover, the list of treatment alternatives, including combination therapies, is rapidly evolving. The molecular profiling and specific tumor-associated immune contexture may be predictive of response or resistance to these therapeutic strategies. Multiplexed immunohistochemistry is an effective and proficient approach to simultaneously identify specific proteins or molecular abnormalities, to determine the spatial distribution and activation state of immune cells, as well as the presence of immunoactive molecular expression. This method is highly advantageous for investigating immune evasion mechanisms and discovering potential biomarkers to assess mechanisms of action and to predict response to a given treatment. This review provides views on the current technological status and evidence for clinical applications of multiplexing and how it could be applied to optimize clinical management of patients with lung cancer.

Practical approaches to automated digital image analysis of Ki-67 labeling index in 997 breast carcinomas and causes of discordance with visual assessment

The Ki-67 labeling index (LI) is an important prognostic factor in breast carcinoma. The Ki-67 LI is traditionally calculated via unaided microscopic estimation; however, inter-observer and intra-observer variability and low reproducibility are problems with this visual assessment (VA) method. For more accurate assessment and better reproducibility with Ki-67 LI, digital image analysis was introduced recently. We used both VA and automated digital image analysis (ADIA) (Ventana Virtuoso image management software) to estimate Ki-67 LI for 997 cases of breast carcinoma, and compared VA and ADIA results. VA and ADIA were highly correlated (intraclass correlation coefficient 0.982, and Spearman’s correlation coefficient 0.966, p<0.05). We retrospectively analyzed cases with a greater than 5% difference between VA and ADIA results. The cause of these differences was: (1) tumor heterogeneity (98 cases, 56.0%), (2) VA interpretation error (32 cases, 18.3%), (3) misidentification of tumor cells (26 cases, 14.9%), (4) poor immunostaining or slide quality (16 cases, 9.1%), and (5) Estimation of non-tumor cells (3 cases, 1.7%). There were more discrepancies between VA and ADIA results in the group with a VA value of 10–20% compared to groups with <10% and ≥20%. Although ADIA is more accurate than VA, there are some limitations. Therefore, ADIA findings require confirmation by a pathologist.

An international multicenter study to evaluate reproducibility of automated scoring for assessment of Ki67 in breast cancer

The nuclear proliferation biomarker Ki67 has potential prognostic, predictive, and monitoring roles in breast cancer. Unacceptable between-laboratory variability has limited its clinical value. The International Ki67 in Breast Cancer Working Group investigated whether Ki67 immunohistochemistry can be analytically validated and standardized across laboratories using automated machine-based scoring. Sets of pre-stained core-cut biopsy sections of 30 breast tumors were circulated to 14 laboratories for scanning and automated assessment of the average and maximum percentage of tumor cells positive for Ki67. Seven unique scanners and 10 software platforms were involved in this study. Pre-specified analyses included evaluation of reproducibility between all laboratories (primary) as well as among those using scanners from a single vendor (secondary). The primary reproducibility metric was intraclass correlation coefficient between laboratories, with success considered to be intraclass correlation coefficient >0.80. Intraclass correlation coefficient for automated average scores across 16 operators was 0.83 (95% credible interval: 0.73-0.91) and intraclass correlation coefficient for maximum scores across 10 operators was 0.63 (95% credible interval: 0.44-0.80). For the laboratories using scanners from a single vendor (8 score sets), intraclass correlation coefficient for average automated scores was 0.89 (95% credible interval: 0.81-0.96), which was similar to the intraclass correlation coefficient of 0.87 (95% credible interval: 0.81-0.93) achieved using these same slides in a prior visual-reading reproducibility study. Automated machine assessment of average Ki67 has the potential to achieve between-laboratory reproducibility similar to that for a rigorously standardized pathologist-based visual assessment of Ki67. The observed intraclass correlation coefficient was worse for maximum compared to average scoring methods, suggesting that maximum score methods may be suboptimal for consistent measurement of proliferation. Automated average scoring methods show promise for assessment of Ki67 scoring, but requires further standardization and subsequent clinical validation.

Ancillary Prognostic and Predictive Testing in Breast Cancer: Focus on Discordant, Unusual, and Borderline Results

Ancillary testing in breast cancer has become standard of care to determine what therapies may be most effective for individual patients with breast cancer. Single-marker tests are required on all newly diagnosed and newly metastatic breast cancers. Markers of proliferation are also used, and include both single-marker tests like Ki67 as well as panel-based gene expression tests, which have made more recent contributions to prognostic and predictive testing in breast cancers. This review focuses on pathologist interpretation of these ancillary test results, with a focus on expected versus unexpected results and troubleshooting borderline, unusual, or discordant results.

Focused scores enable reliable discrimination of small differences in steatosis

Background

Automated image analysis enables quantitative measurement of steatosis in histological images. However, spatial heterogeneity of steatosis can make quantitative steatosis scores unreliable. To improve the reliability, we have developed novel scores that are “focused” on steatotic tissue areas.

Methods

Focused scores use concepts of tile-based hotspot analysis in order to compute statistics about steatotic tissue areas in an objective way. We evaluated focused scores on three data sets of images of rodent liver sections exhibiting different amounts of dietary-induced steatosis. The same evaluation was conducted with the standard steatosis score computed by most image analysis methods.

Results

The standard score reliably discriminated large differences in steatosis (intraclass correlation coefficient ICC = 0.86), but failed to discriminate small (ICC = 0.54) and very small (ICC = 0.14) differences. With an appropriate tile size, mean-based focused scores reliably discriminated large (ICC = 0.92), small (ICC = 0.86) and very small (ICC = 0.83) differences. Focused scores based on high percentiles showed promise in further improving the discrimination of very small differences (ICC = 0.93).

Conclusions

Focused scores enable reliable discrimination of small differences in steatosis in histological images. They are conceptually simple and straightforward to use in research studies.

Programmed death-ligand 1 expression by digital image analysis advances thyroid cancer diagnosis among encapsulated follicular lesions

Recognition of noninvasive follicular thyroid neoplasms with papillary-like nuclear features (NIFTP) that distinguishes them from invasive malignant encapsulated follicular variant of papillary thyroid carcinoma (EFVPTC) can prevent overtreatment of NIFTP patients. We and others have previously reported that programmed death-ligand 1 (PD-L1) is a useful biomarker in thyroid tumors; however, all reports to date have relied on manual scoring that is time consuming as well as subject to individual bias. Consequently, we developed a digital image analysis (DIA) protocol for cytoplasmic and membranous stain quantitation (ThyApp) and evaluated three tumor sampling methods [Systemic Uniform Random Sampling, hotspot nucleus, and hotspot nucleus/3,3'-Diaminobenzidine (DAB)]. A patient cohort of 153 cases consisting of 48 NIFTP, 44 EFVPTC, 26 benign nodules and 35 encapsulated follicular lesions/neoplasms with lymphocytic thyroiditis (LT) was studied. ThyApp quantitation of PD-L1 expression revealed a significant difference between invasive EFVPTC and NIFTP; but none between NIFTP and benign nodules. ThyApp integrated with hotspot nucleus tumor sampling method demonstrated to be most clinically relevant, consumed least processing time, and eliminated interobserver variance. In conclusion, the fully automatic DIA algorithm developed using a histomorphological approach objectively quantitated PD-L1 expression in encapsulated thyroid neoplasms and outperformed manual scoring in reproducibility and higher efficiency.

Using computer assisted image analysis to determine the optimal Ki67 threshold for predicting outcome of invasive breast cancer

Background

Ki67 positivity in invasive breast cancers has an inverse correlation with survival outcomes and serves as an immunohistochemical surrogate for molecular subtyping of breast cancer, particularly ER positive breast cancer. The optimal threshold of Ki67 in both settings, however, remains elusive. We use computer assisted image analysis (CAIA) to determine the optimal threshold for Ki67 in predicting survival outcomes and differentiating luminal B from luminal A breast cancers.

Methods

Quantitative scoring of Ki67 on tissue microarray (TMA) sections of 440 invasive breast cancers was performed using Aperio ePathology ImmunoHistochemistry Nuclear Image Analysis algorithm, with TMA slides digitally scanned via Aperio ScanScope XT System.

Results

On multivariate analysis, tumours with Ki67 ≥14% had an increased likelihood of recurrence (HR 1.941, p=0.021) and shorter overall survival (HR 2.201, p=0.016). Similar findings were observed in the subset of 343 ER positive breast cancers (HR 2.409, p=0.012 and HR 2.787, p=0.012 respectively). The value of Ki67 associated with ER+HER2-PR<20% tumours (Luminal B subtype) was found to be <17%.

Conclusion

Using CAIA, we found optimal thresholds for Ki67 that predict a poorer prognosis and an association with the Luminal B subtype of breast cancer. Further investigation and validation of these thresholds are recommended.

Digital image analysis of Ki67 proliferation index in breast cancer using virtual dual staining on whole tissue sections: clinical validation and inter-platform agreement

Purpose

The Ki67 proliferation index is a prognostic and predictive marker in breast cancer. Manual scoring is prone to inter- and intra-observer variability. The aims of this study were to clinically validate digital image analysis (DIA) of Ki67 using virtual dual staining (VDS) on whole tissue sections and to assess inter-platform agreement between two independent DIA platforms.

Methods

Serial whole tissue sections of 154 consecutive invasive breast carcinomas were stained for Ki67 and cytokeratin 8/18 with immunohistochemistry in a clinical setting. Ki67 proliferation index was determined using two independent DIA platforms, implementing VDS to identify tumor tissue. Manual Ki67 score was determined using a standardized manual counting protocol. Inter-observer agreement between manual and DIA scores and inter-platform agreement between both DIA platforms were determined and calculated using Spearman’s correlation coefficients. Correlations and agreement were assessed with scatterplots and Bland–Altman plots.

Results

Spearman’s correlation coefficients were 0.94 (p < 0.001) for inter-observer agreement between manual counting and platform A, 0.93 (p < 0.001) between manual counting and platform B, and 0.96 (p < 0.001) for inter-platform agreement. Scatterplots and Bland–Altman plots revealed no skewness within specific data ranges. In the few cases with ≥ 10% difference between manual counting and DIA, results by both platforms were similar.

Conclusions

DIA using VDS is an accurate method to determine the Ki67 proliferation index in breast cancer, as an alternative to manual scoring of whole sections in clinical practice. Inter-platform agreement between two different DIA platforms was excellent, suggesting vendor-independent clinical implementability.

Electronic supplementary material

The online version of this article (10.1007/s10549-018-4669-2) contains supplementary material, which is available to authorized users.

Markers of clinical utility in the differential diagnosis and prognosis of prostate cancer

Molecular diagnostics is a rapidly evolving area of surgical pathology, that is gradually beginning to transform our diagnostical procedures for a variety of tumors. Next to molecular prognostication that has begun to complement our histological diagnosis in breast cancer, additional testing to detect targets and to predict therapy response has become common practice in breast and lung cancer. Prostate cancer is a bit slower in this respect, as it is still largely diagnosed and classified on morphological grounds. Our diagnostic immunohistochemical armamentarium of basal cell markers and positive markers of malignancy now allows to clarify the majority of lesions, if applied to the appropriate morphological context (and step sections). Prognostic immunohistochemistry remains a problematic and erratic yet tempting research field that provides information on tumor relevance of proteins, but little hard data to integrate into our diagnostic workflow. Main reasons are various issues of standardization that hamper the reproducibility of cut-off values to delineate risk categories. Molecular testing of DNA-methylation or transcript profiling may be much better standardized and this review discusses a couple of commercially available tests: The ConfirmDX test measures DNA-methylation to estimate the likelihood of cancer detection on a repeat biopsy and may help to reduce unnecessary biopsies. The tests Prolaris, OncotypeDX Prostate, and Decipher all are transcript tests that have shown to provide prognostic data independent of clinico-pathological parameters and that may aid in therapy planning. However, further validation and more comparative studies will be needed to clarify the many open questions concerning sampling bias and tumor heterogeneity.

Interlaboratory variability of Ki67 staining in breast cancer

BACKGROUND

Postanalytic issues of Ki67 assessment in breast cancers like counting method standardisation and interrater bias have been subject of various studies, but little is known about analytic variability of Ki67 staining between pathology labs. Our aim was to study interlaboratory variability of Ki67 staining in breast cancer using tissue microarrays (TMAs) and central assessment to minimise preanalytic and postanalytic influences.

METHODS

Thirty European pathology labs stained serial slides of a TMA set of breast cancer tissues with Ki67 according to their routine in-house protocol. The Ki67-labelling index (Ki67-LI) of 70 matched samples was centrally assessed by one observer who counted all cancer cells per sample. We then tested for differences between the labs in Ki67-LI medians by analysing variance on ranks and in proportions of tumours classified as luminal A after dichotomising oestrogen receptor-positive cancers into cancers showing low (<14%, luminal A) and high (≥14%, luminal B HER2 negative) Ki67-LI using Cochran's Q.

RESULTS

Substantial differences between the 30 labs were indicated for median Ki67-LI (0.65%-33.0%, p < 0.0001) and proportion of cancers classified as luminal A (17%-57%, p < 0.0001). The differences remained significant when labs using the same antibody (MIB-1, SP6, or 30-9) were analysed separately or labs without prior participation in external quality assurance programs were excluded (p < 0.0001, respectively).

CONCLUSION

Substantial variability in Ki67 staining of breast cancer tissue was found between 30 routine pathology labs. Clinical use of the Ki67-LI for therapeutic decisions should be considered only fully aware of lab-specific reference values.

[Histological grading of breast cancer]

From a historical perspective, histological grading was the earliest cell-based method for assessing tumor biology and the prognosis of breast cancer. This review article provides detailed and practical instructions for grading of breast cancer in routine diagnostics. Furthermore, the increasing relevance of precise histological grading in the era of molecular pathology is discussed.

St Gallen 2015 subtyping of luminal breast cancers: impact of different Ki67-based proliferation assessment methods

Ki67 has been proposed as prognostic proliferation marker in luminal breast cancer (BC), but little is known on the influence of Ki67 assessment methods on subtyping into luminal A- and B-like tumors. Our aim was to study the influence of different Ki67-labeling index (Ki67-LI) assessment methods on the proportion of BCs classified as luminal A-like. 280 early BCs were subtyped according to the St Gallen 2015 definitions into 71 % luminal (HER2 negative), 6 % luminal B-like (HER2 positive), 13 % triple negative, 1 % HER2 positive (nonluminal), and 9 % special type. Digitized whole slides were counted manually on the screen. We used nine defined counting methods to assess the Ki67-LI (including the International Ki67 in Breast Cancer Working Group recommendations), and compared the resulting medians and the proportions of cancers classified as luminal A-like according to the formerly used cut-off <20 %. Methods assessing hot spots and tumor periphery resulted in significantly higher Ki67-LI medians than those measuring an average proliferation (27.45 % vs 16.96 %, p < 0.0001). Substantially lower median Ki67-LI were found when assessing 1020 compared to counting 100, 200, 300 cells (17.65 vs 33 %, vs 28 %, vs 24.33 %, respectively; p < 0.0001), or 510 cells (20.59 %, p = 0.019). Applying a standard Ki67-LI cut-off <20 % to define low proliferation for all methods, the proportion of luminal A-like cancers varied between 13 and 44 %. The proportion of BCs classified as luminal A-like is highly influenced by the Ki67-LI assessment method. As a consequence, the selection of a specific Ki67-LI assessment method may have a direct effect on the proportion of patients considered having low-risk disease and thus influence therapeutic decision making. This calls for a standardized assessment method.

Intratumoral heterogeneity of Ki67 expression in early breast cancers exceeds variability between individual tumours

AIMS

Regional differences in proliferative activity are commonly seen within breast cancers, but little is known on the extent of intratumoral heterogeneity of Ki67 expression. Our aim was to study the intratumoral heterogeneity of Ki67 expression in early breast cancers and its association with clinicopathological features, such as oestrogen receptor (ER) status, grade and histological subtype.

METHODS AND RESULTS

The Ki67-labelling index (Ki67-LI) was assessed in hot, cold and intermediate spots of 233 invasive breast cancers by counting a total of 1020 cells, according to a protocol of the International Ki67 in Breast Cancer Working Group. Differences between the spots per tumour were analysed further for clinicopathological subgroups defined by ER status, grade and histological subtype. All clinicopathological subgroups showed significant differences in Ki67-LI between hot, intermediate and cold spots (P < 0.0001). The coefficient of variance (CV) between the spots was higher in ER-positive than in ER-negative cancers (72.6 versus 49.2%, P < 0.0001), and was highest in grade 3 (96.12%), grade 1 (87.27%) and invasive lobular tumours (83.59%) and lowest in medullary (26.48%) cancers. Nested analysis of variance indicated that in both ER-positive and ER-negative cancers, variance in Ki67-LI within tumours contributed more to the total variance (56% for ER-positive, 60% for ER-negative cancers) than the variance between tumours.

CONCLUSION

Intratumoral heterogeneity in Ki67-LI is a ubiquitous phenomenon across various pathological subgroups of breast cancer that may impact assessment of Ki67 levels for clinical decision-making, and sheds new light on recommended cut-offs.

Digital image analysis outperforms manual biomarker assessment in breast cancer

In the spectrum of breast cancers, categorization according to the four gene expression-based subtypes 'Luminal A,' 'Luminal B,' 'HER2-enriched,' and 'Basal-like' is the method of choice for prognostic and predictive value. As gene expression assays are not yet universally available, routine immunohistochemical stains act as surrogate markers for these subtypes. Thus, congruence of surrogate markers and gene expression tests is of utmost importance. In this study, 3 cohorts of primary breast cancer specimens (total n=436) with up to 28 years of survival data were scored for Ki67, ER, PR, and HER2 status manually and by digital image analysis (DIA). The results were then compared for sensitivity and specificity for the Luminal B subtype, concordance to PAM50 assays in subtype classification and prognostic power. The DIA system used was the Visiopharm Integrator System. DIA outperformed manual scoring in terms of sensitivity and specificity for the Luminal B subtype, widely considered the most challenging distinction in surrogate subclassification, and produced slightly better concordance and Cohen's κ agreement with PAM50 gene expression assays. Manual biomarker scores and DIA essentially matched each other for Cox regression hazard ratios for all-cause mortality. When the Nottingham combined histologic grade (Elston-Ellis) was used as a prognostic surrogate, stronger Spearman's rank-order correlations were produced by DIA. Prognostic value of Ki67 scores in terms of likelihood ratio χ(2) (LR χ(2)) was higher for DIA that also added significantly more prognostic information to the manual scores (LR-Δχ(2)). In conclusion, the system for DIA evaluated here was in most aspects a superior alternative to manual biomarker scoring. It also has the potential to reduce time consumption for pathologists, as many of the steps in the workflow are either automatic or feasible to manage without pathological expertise.

Bimodality of intratumor Ki67 expression is an independent prognostic factor of overall survival in patients with invasive breast carcinoma

Proliferative activity, assessed by Ki67 immunohistochemistry (IHC), is an established prognostic and predictive biomarker of breast cancer (BC). However, it remains under-utilized due to lack of standardized robust measurement methodologies and significant intratumor heterogeneity of expression. A recently proposed methodology for IHC biomarker assessment in whole slide images (WSI), based on systematic subsampling of tissue information extracted by digital image analysis (DIA) into hexagonal tiling arrays, enables computation of a comprehensive set of Ki67 indicators, including intratumor variability. In this study, the tiling methodology was applied to assess Ki67 expression in WSI of 152 surgically removed Ki67-stained (on full-face sections) BC specimens and to test which, if any, Ki67 indicators can predict overall survival (OS). Visual Ki67 IHC estimates and conventional clinico-pathologic parameters were also included in the study. Analysis revealed linearly independent intrinsic factors of the Ki67 IHC variance: proliferation (level of expression), disordered texture (entropy), tumor size and Nottingham Prognostic Index, bimodality, and correlation. All visual and DIA-generated indicators of the level of Ki67 expression provided significant cutoff values as single predictors of OS. However, only bimodality indicators (Ashman's D, in particular) were independent predictors of OS in the context of hormone receptor and HER2 status. From this, we conclude that spatial heterogeneity of proliferative tumor activity, measured by DIA of Ki67 IHC expression and analyzed by the hexagonal tiling approach, can serve as an independent prognostic indicator of OS in BC patients that outperforms the prognostic power of the level of proliferative activity.

A methodology for comprehensive breast cancer Ki67 labeling index with intra-tumor heterogeneity appraisal based on hexagonal tiling of digital image analysis data

Digital image analysis (DIA) enables higher accuracy, reproducibility, and capacity to enumerate cell populations by immunohistochemistry; however, the most unique benefits may be obtained by evaluating the spatial distribution and intra-tissue variance of markers. The proliferative activity of breast cancer tissue, estimated by the Ki67 labeling index (Ki67 LI), is a prognostic and predictive biomarker requiring robust measurement methodologies. We performed DIA on whole-slide images (WSI) of 302 surgically removed Ki67-stained breast cancer specimens; the tumour classifier algorithm was used to automatically detect tumour tissue but was not trained to distinguish between invasive and non-invasive carcinoma cells. The WSI DIA-generated data were subsampled by hexagonal tiling (HexT). Distribution and texture parameters were compared to conventional WSI DIA and pathology report data. Factor analysis of the data set, including total numbers of tumor cells, the Ki67 LI and Ki67 distribution, and texture indicators, extracted 4 factors, identified as entropy, proliferation, bimodality, and cellularity. The factor scores were further utilized in cluster analysis, outlining subcategories of heterogeneous tumors with predominant entropy, bimodality, or both at different levels of proliferative activity. The methodology also allowed the visualization of Ki67 LI heterogeneity in tumors and the automated detection and quantitative evaluation of Ki67 hotspots, based on the upper quintile of the HexT data, conceptualized as the "Pareto hotspot". We conclude that systematic subsampling of DIA-generated data into HexT enables comprehensive Ki67 LI analysis that reflects aspects of intra-tumor heterogeneity and may serve as a methodology to improve digital immunohistochemistry in general.