Machine learning evaluation of immune infiltrate through digital tumour score allows prediction of survival outcome in a pooled analysis of three international stage III colon cancer cohorts.

BACKGROUND: T-cell immune infiltrates are robust prognostic variables in localised colon cancer. Evaluation of prognosis using artificial intelligence is an emerging field. We evaluated whether machine learning analysis improved prediction of patient outcome in comparison with analysis of T cell infiltrate only or in association with clinical variables. METHODS: We used data from two phase III clinical trials (Prodige-13 and PETACC08) and one retrospective Italian cohort (HARMONY). Cohorts were split into training (N = 692), internal validation (N = 297) and external validation (N = 672) sets. Tumour slides were stained with CD3mAb. CD3 Machine Learning (CD3ML) score was computed using graphical parameters within the tumour tiles obtained from CD3 slides. CD3 infiltrates in tumour core and invasive margin were automatically detected. Associations of CD3 infiltrates and CD3ML with 5-year Disease-Free Survival (DFS) were examined using univariate and multivariable survival models by Cox regression. FINDINGS: CD3 density both in the invasive margin and the tumour core were significantly associated with DFS in the different sets. Similarly, CD3ML score was significantly associated with DFS in all sets. CD3 assessment did not provide added value on top of CD3ML assessment (Likelihood Ratio Test (LRT), p = 0.13). In contrast, CD3ML improved prediction of DFS when combined with a clinical risk stage (LRT, p = 0.001). Stratified by clinical risk score (High or Low), patients with low CD3ML score had better DFS. INTERPRETATION: In all tested sets, machine learning analysis of tumour cells improved prediction of prognosis compared to clinical parameters. Adding tumour-infiltrating lymphocytes assessment did not improve prognostic determination. FUNDING: This research received no external funding.

CD3-CD8 immune score associated with a clinical score stratifies PDAC prognosis regardless of adjuvant or neoadjuvant chemotherapy.

Stratification of the prognosis of pancreatic cancer (PDAC) patients treated by surgery is based solely on clinical variables, such as tumor stage and node status. The development of biomarkers of relapse is needed, especially to drive administration of adjuvant therapy in this at-risk population. Our study evaluates the prognostic performance of a CD3- and CD8-based immune score. CD3, CD8 and Foxp3 expression were evaluated on whole slides in two retrospective PDAC cohorts totaling 334 patients. For this study, we developed an immune score to estimate CD3 and CD8 infiltration in both tumor core and invasive margin using computer-guided analysis with QuPath software. Variables were combined in a dichotomous immune score. The association between immune and clinical scores, and both PFS and OS was investigated. We observed that a dichotomous immune score predicts both PFS and OS of localized PDAC. By univariate and multivariate analysis, immune score, tumor grade, adjuvant therapy, lymph node status, and adjuvant chemotherapy administration were associated with PFS and OS. We subsequently associated the PDAC immune score and clinical variables in a combined score. This combined score predicted patient outcomes independently of adjuvant or neoadjuvant treatment, and improved patient prognostic prediction compared to clinical variables or immune score alone.

Pathologic complete response and survival in HER2-low and HER2-zero early breast cancer treated with neoadjuvant chemotherapy.

BACKGROUND: Breast cancers without HER2 amplification but still expressing this membrane protein constitute a new entity called HER2-low tumors. It is important to characterize them in terms of sensitivity to treatment and prognosis. PATIENTS AND METHODS: To investigate chemosensitivity and long-term prognosis of HER2-low early breast cancer (eBC), compared to HER2-0 tumors, we retrospectively retrieved clinicopathological characteristics, response to treatment, and survival data from 511 patients treated for eBC with neoadjuvant chemotherapy (NAC) in a French cancer center between 2007 and 2018. Factors associated with the achievement of pathologic complete response (pCR) and survival were studied among hormone receptor positive (HR+) and negative (HR-) eBC. RESULTS: A total of 280 HR+ (61% HER2-low), and 231 HR- (28% HER2-low) eBC were included. We found classical clinicopathological factors usually associated with chemosensitivity and prognosis, in both HR+ and HR- eBC. By uni- and multivariable analysis, HER2 status (low vs 0) was not independently associated with pCR, either in HR+ or HR- eBC. Relapse free (RFS) and overall survival (OS) were not significantly different between HER2-low and HER2-0 among HR+ tumors. In contrast, among HR- negative tumors, RFS and OS were slightly better in HER2-0 eBC by univariable but not by multivariable analysis. CONCLUSIONS: In eBC patients treated with NAC, taking into account HR expression subtype and other current clinicopathological features, HER2-low tumors did not appear to have different chemosensitivity or prognosis, compared to their HER2-0 counterparts.

Immunoscore immune checkpoint using spatial quantitative analysis of CD8 and PD-L1 markers is predictive of the efficacy of anti- PD1/PD-L1 immunotherapy in non-small cell lung cancer.

BACKGROUND: Anti-PD-1 and PD-L1 antibodies (mAbs) are approved immunotherapy agents to treat metastatic non-small cell lung cancer (NSCLC) patients. Only a minority of patients responds to these treatments and biomarkers predicting response are currently lacking. METHODS: Immunoscore-Immune-Checkpoint (Immunoscore-IC), an in vitro diagnostic test, was used on 471 routine single FFPE-slides, and the duplex-immunohistochemistry CD8 and PD-L1 staining was quantified using digital-pathology. Analytical validation was performed on two independent cohorts of 206 NSCLC patients. Quantitative parameters related to cell location, number, proximity and clustering were analysed. The Immunoscore-IC was applied on a first cohort of metastatic NSCLC patients (n = 133), treated with anti-PD1 or anti-PD-L1 mAbs. Another independent cohort (n = 132) served as validation. FINDINGS: Anti-PDL1 clone (HDX3) has similar characteristics as anti-PD-L1 clones (22C3, SP263). Densities of PD-L1+ cells, CD8+ cells and distances between CD8+ and PD-L1+ cells were quantified and the Immunoscore-IC classification was computed. Using univariate Cox model, 5 histological dichotomised variables (CD8 free of PD-L1+ cells, CD8 clusters, CD8 cells in proximity of PD-L1 cells, CD8 density and PD-L1 cells in proximity of CD8 cells) were significantly associated with Progression-Free Survival (PFS) (all P < 0.0001). Immunoscore-IC classification improved the discriminating power of prognostic model, which included clinical variables and pathologist PD-L1 assessment. In two categories, the Immunoscore-IC risk-score was significantly associated with patients' PFS (HR = 0.39, 95% CI (0.26-0.59), P < 0.0001) and Overall Survival (OS) (HR = 0.42, 95% CI (0.27-0.65), P < 0.0001) in the training-set. Further increased hazard ratios (HR) were found when stratifying patients into three-category Immunoscore-IC (IS-IC). All patients with Low-IS-IC progressed in less than 18 months, whereas PFS at 36 months were 34% and 33% of High-IS-IC patients in the training and validation sets, respectively. INTERPRETATION: Immunoscore-IC is a powerful tool to predict the efficacy of immune-checkpoint inhibitors (ICIs) in patients with NSCLC. FUNDING: Veracyte, INSERM, Labex Immuno-Oncology, Transcan ERAnet European project, ARC, SIRIC, CARPEM, Ligue Contre le Cancer, ANR, QNRF, INCa France, Louis Jeantet Prize Foundation.

PIK3CA and PIK3R1 tumor mutational landscape in a pan-cancer patient cohort and its association with pathway activation and treatment efficacy.

There is little data concerning the implications of PIK3CA mutations outside of the known hotspots described in ER+/HER2- metastatic breast cancer (mBC). Similarly, PIK3R1 mutations could also lead to activation of PI3K pathway, but are poorly described. We determined the incidence and type of all somatic PIK3CA and PIK3R1 mutations by whole exome sequencing (WES) in a pan-cancer cohort of 1200 patients. Activation of the PI3K pathway was studied using phospho-AKT immunohistochemistry. Associations between PIK3CA/PIK3R1 mutations and response to chemotherapy were studied in mBC cases. We found 141 patients (11.8%) with a PIK3CA and/or PIK3R1 mutation across 20 different cancer types. The main cancer subtype was mBC (45.4%). Eighty-four mutations (62.2%) occurred in the three described hotspots; 51 mutations occurred outside of these hotspots. In total, 78.4% were considered activating or probably activating. Among PIK3R1 mutations, 20% were loss of function mutations, leading to a constitutional activation of the pathway. Phospho-AKT quantification in tumor samples was in favor of activation of the PI3K pathway in the majority of mutated tumors, regardless of mutation type. In ER+/HER2- mBC, first line chemotherapy efficacy was similar for PIK3CA-mutated and PIK3CA-WT tumors, whereas in triple negative mBC, chemotherapy appeared to be more effective in PIK3CA-WT tumors. In this large, real-life pan-cancer patient cohort, our results indicate that PIK3CA/PIK3R1 mutations are widely spread, and plead in favour of evaluating the efficacy of PI3K inhibitors outside of ER+/HER2- mBC and outside of hotspot mutations.

Pathologic and immunohistochemical prognostic markers in residual triple-negative breast cancer after neoadjuvant chemotherapy.

Background: The persistence of residual tumour after neoadjuvant chemotherapy (NAC) in localised triple-negative breast cancer (TNBC) is known to have a negative prognostic value. However, different degrees of expression of some immunohistochemical markers may correlate with different prognoses. Methods: The expression of biomarkers with a known prognostic value, i.e., cytokeratin 5/6 (CK5/6), androgen receptor (AR), epidermal growth factor receptor (EGFR) proliferation-related nuclear antigen Ki-67, human epidermal growth factor receptor 2 (HER2), protein 53 (p53), forkhead box protein 3 (FOXP3), and cluster differentiation 8 (CD8), was analysed by immunohistochemistry in 111 samples after NAC in non-metastatic TNBC patients addressed to Georges-François Leclerc Cancer Centre Dijon, France. Clinical and pathological variables were retrospectively collected. Cox regression was used to identify immunohistochemical (IHC) and clinicopathological predictors of event-free survival (EFS) (relapse or death). Results: Median age was 50.4 years (range 25.6-88.3), 55.9% (n = 62) were non-menopausal, 70 (63.1%) had stage IIA-IIB disease. NAC was mostly sequential anthracycline-taxanes (72.1%), and surgical intervention was principally conservative (51.3%). We found 65.7% ypT1, 47.2% lymph node involvement (ypN+), and 29.4% lymphovascular invasion (LVI). Most residual tumours were EGFR >110 (H-score) (60.5%, n = 66), AR ≥4% (53.2%, n = 58), p53-positive mutated (52.7%, n = 58), CD8 ≥26 (58.1%, n = 61), FOXP3 ≥7 (51.4%, n = 54), more than half in the stroma, and 52.3% (n = 58) HER2 score 0. After a median follow-up of 80.8 months, 48.6% had relapsed. Median EFS was 62.3 months (95% CI, 37.2-not reached (NR)). Factors independently associated with poor EFS were AR-low (p = 0.002), ypN+ (p < 0.001), and LVI (p = 0.001). Factors associated with lower overall survival (OS) were EGFR-low (p = 0.041), Ki-67 high (p = 0.024), and ypN+ (p < 0.001). Conclusion: Post-NAC residual disease in TNBC showed biomarkers specific to a basal-like subtype and markers of lymphocyte infiltration mostly present in the stroma. Prognostic markers for EFS were AR, LVI, and ypN and warrant further validation in a prognostic model.

Combination of CDX2 H-score quantitative analysis with CD3 AI-guided analysis identifies patients with a good prognosis only in stage III colon cancer.

AIM: Stratification of colon cancer (CC) of patients with stage II and III for risk of relapse is still needed especially to drive adjuvant therapy administration. Our study evaluates the prognostic performance of two known biomarkers, CDX2 and CD3, standalone or their combined information in stage II and III CC. PATIENTS AND METHODS: CDX2 and CD3 expression was evaluated in Prodige-13 study gathering 443 stage II and 398 stage III primary CC on whole slide colectomy. We developed for this study an H-score to quantify CDX2 expression and used our artificial intelligence (AI)-guided tissue analysis ColoClass to detect CD3 in tumour core and invasive margin. Association between biomarkers and relapse-free survival was investigated. RESULTS: Univariate analysis showed that the combined variable CD3-TC and CD3-IM was associated with prognosis in both stage II and stage III. CDX2, on the contrary, was associated with prognosis only in stage III. We subsequently associated CDX2 and combined immune parameters only in stage III. This multivariate analysis allowed us to distinguish a proportion of stage III CC harbouring a high CDX2 expression and a high immune infiltration with a particularly good prognosis compared to their counterpart. CONCLUSION: This study validated the prognostic role of CDX2 and CD3 evaluated with immunohistochemistry procedures in stage III but not in stage II. This association would be conceivable in a routine pathology laboratory and could help oncologist to consider chemotherapy de-escalation for a part of stage III patients.

Using a convolutional neural network for classification of squamous and non-squamous non-small cell lung cancer based on diagnostic histopathology HES images.

Histological stratification in metastatic non-small cell lung cancer (NSCLC) is essential to properly guide therapy. Morphological evaluation remains the basis for subtyping and is completed by additional immunohistochemistry labelling to confirm the diagnosis, which delays molecular analysis and utilises precious sample. Therefore, we tested the capacity of convolutional neural networks (CNNs) to classify NSCLC based on pathologic HES diagnostic biopsies. The model was estimated with a learning cohort of 132 NSCLC patients and validated on an external validation cohort of 65 NSCLC patients. Based on image patches, a CNN using InceptionV3 architecture was trained and optimized to classify NSCLC between squamous and non-squamous subtypes. Accuracies of 0.99, 0.87, 0.85, 0.85 was reached in the training, validation and test sets and in the external validation cohort. At the patient level, the CNN model showed a capacity to predict the tumour histology with accuracy of 0.73 and 0.78 in the learning and external validation cohorts respectively. Selecting tumour area using virtual tissue micro-array improved prediction, with accuracy of 0.82 in the external validation cohort. This study underlines the capacity of CNN to predict NSCLC subtype with good accuracy and to be applied to small pathologic samples without annotation.

Heat shock and HSP70 regulate 5-FU-mediated caspase-1 activation in myeloid-derived suppressor cells and tumor growth in mice.

BACKGROUND: We have previously shown that 5-fluorouracil (5-FU) selectively kills myeloid-derived suppressor cells (MDSCs) and activates NLRP3 (NOD-leucine rich repeat and pyrin containing protein 3) inflammasome. NLRP3 activation leads to caspase-1 activation and production of IL-1ß, which in turn favors secondary tumor growth. We decided to explore the effects of either a heat shock (HS) or the deficiency in heat shock protein (HSP) 70, previously shown to respectively inhibit or increase NLRP3 inflammasome activation in macrophages. METHODS: Caspase-1 activation was detected in vitro in MSC-2 cells by western blot and in vivo or ex vivo in tumor and/or splenic MDSCs by flow cytometry. The effects of HS, HSP70 deficiency and anakinra (an IL-1 inhibitor) on tumor growth and mice survival were studied in C57BL/6 WT or Hsp70-/- tumor-bearing mice. Finally, Th17 polarization was evaluated by qPCR (Il17a, Rorc) and angiogenic markers by qPCR (Pecam1, Eng) and immunohistochemistry (ERG). RESULTS: HS inhibits 5-FU-mediated caspase-1 activation in vitro and in vivo without affecting its cytotoxicity on MDSCs. Moreover, it enhances the antitumor effect of 5-FU treatment and favors mice survival. Interestingly, it is associated to a decreased Th17 and angiogenesis markers in tumors. IL-1ß injection is able to bypass HS+5-FU antitumor effects. In contrast, in Hsp70-/- MDSCs, 5-FU-mediated caspase-1 activation is increased in vivo and in vitro without effect on 5-FU cytotoxicity. In Hsp70-/- mice, the antitumor effect of 5-FU was impeded, with an increased Th17 and angiogenesis markers in tumors. Finally, the effects of 5-FU on tumor growth can be restored by inhibiting IL-1ß, using anakinra. CONCLUSION: This study provides evidence on the role of HSP70 in tuning 5-FU antitumor effect and suggests that HS can be used to improve 5-FU anticancer effect.

Pathologic Response After Weekly Paclitaxel versus Docetaxel in Operable Breast Cancer.

INTRODUCTION: Weekly paclitaxel (Ptx) and q3w docetaxel (Dtx) are equivalent in adjuvant breast cancer treatment. Weekly Ptx is better tolerated than q3w Dtx and became the first choice in daily practice, even preoperatively. METHODS: To compare the efficacy and safety of the two regimens, a retrospective analysis was performed in breast cancer patients (pts) referred for neoadjuvant, sequential, taxane-containing chemotherapy to the Institute of Oncology and Oncofort Clinic, Bucharest, between 2008 and 2017. RESULTS: Forty-seven cases were eligible, median age was 56 years (34-73 years), mainly stage IIIA-B (53.2%, 25 pts) and ductal invasive (70.2%, 33 pts) of which 24 pts (51%) received q3w Dtx and 23 pts (48.9%) weekly Ptx. The histological response rates were 62.5% (15 pts) and 73.7% (17 pts) (p=0.47), average dose-intensity was 87.7% and 96.7% (p=0.002) and grade III-IV toxicity rate was 12.5% and 13% (p=0.64), respectively. Pathologic response was correlated with immunophenotype, PgR expression, tumor size and backbone chemotherapy (p<0.05). DISCUSSION: Our study showed an improved efficacy of taxane's weekly administration, probably due to a better tolerance and a lower rate of dose-impairing toxicities.