Use of Deep Learning to Evaluate Tumor Microenvironmental Features for Prediction of Colon Cancer Recurrence.

Deep learning may detect biologically important signals embedded in tumor morphologic features that confer distinct prognoses. Tumor morphologic features were quantified to enhance patient risk stratification within DNA mismatch repair (MMR) groups using deep learning. Using a quantitative segmentation algorithm (QuantCRC) that identifies 15 distinct morphologic features, we analyzed 402 resected stage III colon carcinomas [191 deficient (d)-MMR; 189 proficient (p)-MMR] from participants in a phase III trial of FOLFOX-based adjuvant chemotherapy. Results were validated in an independent cohort (176 d-MMR; 1,094 p-MMR). Association of morphologic features with clinicopathologic variables, MMR, KRAS, BRAFV600E, and time-to-recurrence (TTR) was determined. Multivariable Cox proportional hazards models were developed to predict TTR. Tumor morphologic features differed significantly by MMR status. Cancers with p-MMR had more immature desmoplastic stroma. Tumors with d-MMR had increased inflammatory stroma, epithelial tumor-infiltrating lymphocytes (TIL), high-grade histology, mucin, and signet ring cells. Stromal subtype did not differ by BRAFV600E or KRAS status. In p-MMR tumors, multivariable analysis identified tumor-stroma ratio (TSR) as the strongest feature associated with TTR [HRadj 2.02; 95% confidence interval (CI), 1.14-3.57; P = 0.018; 3-year recurrence: 40.2% vs. 20.4%; Q1 vs. Q2-4]. Among d-MMR tumors, extent of inflammatory stroma (continuous HRadj 0.98; 95% CI, 0.96-0.99; P = 0.028; 3-year recurrence: 13.3% vs. 33.4%, Q4 vs. Q1) and N stage were the most robust prognostically. Association of TSR with TTR was independently validated. In conclusion, QuantCRC can quantify morphologic differences within MMR groups in routine tumor sections to determine their relative contributions to patient prognosis, and may elucidate relevant pathophysiologic mechanisms driving prognosis. SIGNIFICANCE: A deep learning algorithm can quantify tumor morphologic features that may reflect underlying mechanisms driving prognosis within MMR groups. TSR was the most robust morphologic feature associated with TTR in p-MMR colon cancers. Extent of inflammatory stroma and N stage were the strongest prognostic features in d-MMR tumors. TIL density was not independently prognostic in either MMR group.

Improved Risk-Stratification Scheme for Mismatch-Repair Proficient Stage II Colorectal Cancers Using the Digital Pathology Biomarker QuantCRC.

PURPOSE: There is a need to improve current risk stratification of stage II colorectal cancer to better inform risk of recurrence and guide adjuvant chemotherapy. We sought to examine whether integration of QuantCRC, a digital pathology biomarker utilizing hematoxylin and eosin-stained slides, provides improved risk stratification over current American Society of Clinical Oncology (ASCO) guidelines. EXPERIMENTAL DESIGN: ASCO and QuantCRC-integrated schemes were applied to a cohort of 398 mismatch-repair proficient (MMRP) stage II colorectal cancers from three large academic medical centers. The ASCO stage II scheme was taken from recent guidelines. The QuantCRC-integrated scheme utilized pT3 versus pT4 and a QuantCRC-derived risk classification. Evaluation of recurrence-free survival (RFS) according to these risk schemes was compared using the log-rank test and HR. RESULTS: Integration of QuantCRC provides improved risk stratification compared with the ASCO scheme for stage II MMRP colorectal cancers. The QuantCRC-integrated scheme placed more stage II tumors in the low-risk group compared with the ASCO scheme (62.5% vs. 42.2%) without compromising excellent 3-year RFS. The QuantCRC-integrated scheme provided larger HR for both intermediate-risk (2.27; 95% CI, 1.32-3.91; P = 0.003) and high-risk (3.27; 95% CI, 1.42-7.55; P = 0.006) groups compared with ASCO intermediate-risk (1.58; 95% CI, 0.87-2.87; P = 0.1) and high-risk (2.24; 95% CI, 1.09-4.62; P = 0.03) groups. The QuantCRC-integrated risk groups remained prognostic in the subgroup of patients that did not receive any adjuvant chemotherapy. CONCLUSIONS: Incorporation of QuantCRC into risk stratification provides a powerful predictor of RFS that has potential to guide subsequent treatment and surveillance for stage II MMRP colorectal cancers.

Tumor Budding Assessment in Colorectal Carcinoma: Normalization Revisited.

Tumor budding (TB) is a powerful prognostic factor in colorectal cancer (CRC). An internationally standardized method for its assessment (International Tumor Budding Consensus Conference [ITBCC] method) has been adopted by most CRC pathology protocols. This method requires that TB counts are reported by field area (0.785 mm 2 ) rather than objective lens and a normalization factor is applied for this purpose. However, the validity of this approach is yet to be tested. We sought to validate the ITBCC method with a particular emphasis on normalization as a tool for standardization. In a cohort of 365 stage I-III CRC, both normalized and non-normalized TB were significantly associated with disease-specific survival and recurrence-free survival ( P <0.0001). Examining both 0.95 and 0.785 mm 2 field areas in a subset of patients (n=200), we found that normalization markedly overcorrects TB counts: Counts obtained in a 0.95 mm 2 hotspot field were reduced by an average of 17.5% following normalization compared with only 3.8% when counts were performed in an actual 0.785 mm 2 field. This resulted in 45 (11.3%) cases being downgraded using ITBCC grading criteria following normalization, compared with only 5 cases (1.3%, P =0.0007) downgraded when a true 0.785 mm 2 field was examined. In summary, the prognostic value of TB was retained regardless of whether TB counts in a 0.95 mm 2 field were normalized. Normalization resulted in overcorrecting TB counts with consequent downgrading of most borderline cases. This has implications for risk stratification and adjuvant treatment decisions, and suggests the need to re-evaluate the role of normalization in TB assessment.

Impact of tissue sampling on detection of venous invasion in colorectal cancer: a prospective analysis.

AIMS: Venous invasion (VI) is a powerful yet under-reported prognostic factor in colorectal cancer (CRC). Efforts to improve its detection have largely focused upon histological assessment, with less attention paid to tissue-sampling strategies. This study aimed to prospectively determine the number of tumour blocks required to optimise VI detection in CRC resections. In addition, the relationship between linear spiculation (LS) and extramural venous invasion (EMVI) was investigated. METHODS AND RESULTS: A standardised tissue sampling protocol was developed and applied prospectively to 217 CRC resections [AJCC 8th edition, stage 1 (n = 32); stage 2 (n = 84); stage 3 (n = 87); stage 4 (n = 14); and post-neoadjuvant therapy (n = 46)]. Elastin stains were performed on all tumour blocks. VI was identified in 55% of cases (EMVI = 37%; IMVI alone = 18%). The sensitivity of VI detection increased with increasing numbers of tumour blocks submitted [one block (35%), three blocks (66%), five blocks (84%), six blocks (95%) and seven blocks (97%)]. Similar findings were observed for EMVI [one block (35%), three blocks (73%), five blocks (89%), six blocks (96%) and seven blocks (96%)]. LS was identified macroscopically in 22% of specimens. In cases where no neoadjuvant therapy had been given, EMVI was significantly associated with LS (71% in LS+ cases versus 29% in LS- cases; P < 0.001). In addition, tumour blocks targeting LS were associated with a fivefold higher rate of EMVI compared with blocks that did not (P < 0.001). CONCLUSIONS: Our findings demonstrate the impact of tissue sampling and quality of gross examination on VI detection and may inform practices in future CRC protocols.

Mast cell evaluation in gastrointestinal biopsies: should we be counting? A critical review and practical guide for the surgical pathologist.

Mast cells are residents of the tubular gastrointestinal (GI) tract, where they play an important role in host defence and other vital functions. Dysregulation of mast cells has been implicated in the pathogenesis of several neoplastic, inflammatory, and functional disorders, some of which may manifest with GI symptoms. Surgical pathologists must therefore confront when and how to evaluate GI biopsies for mast cells, and whether such decisions should be based on morphologic criteria, clinical context, or direct request from clinical colleagues. The pathologist's role in evaluation of mast cell infiltrates is best defined in the diagnosis of systemic mastocytosis, where the utility of morphologic assessment coupled with ancillary studies is well established. In contrast, in nonneoplastic mast cell disorders such as mast cell activation syndrome, irritable bowel syndrome, or so-called 'mastocytic enterocolitis', a role for histopathology, if any, is controversial. Despite this, pathologists have seen a sharp increase in requests for mast cell quantification in the latter setting, despite these requests not being supported by published evidence. Moreover, what constitutes a 'normal' number of mast cells in a luminal GI biopsy is not well established. As a result, there is considerable variation in how these requests are handled in practice. This review evaluates and summarizes the published evidence relating to mast cell evaluation in endoscopic GI biopsies in various clinical scenarios, with a goal of providing practical, evidence-based guidance for the surgical pathologist when approached with requests for mast cell quantification in GI biopsies.

Resident Depression and Burnout During the COVID-19 Pandemic: A Survey of Canadian Laboratory Medicine Trainees.

CONTEXT.­: Resident physicians face a higher rate of burnout and depression than the general population. Few studies have examined burnout and depression in Canadian laboratory medicine residents, and none during the COVID-19 pandemic. OBJECTIVE.­: To identify the prevalence of burnout and depression, contributing factors, and the impact of COVID-19 in this population. DESIGN.­: An electronic survey was distributed to Canadian laboratory medicine residents. Burnout was assessed using the Oldenburg Burnout Inventory. Depression was assessed using the Patient Health Questionnaire 9. RESULTS.­: Seventy-nine responses were collected. The prevalence of burnout was 63% (50 of 79). The prevalence of depression was 47% (37 of 79). Modifiable factors significantly associated with burnout included career dissatisfaction, below average academic performance, lack of time off for illness, stress related to finances, lack of a peer or staff physician mentor, and a high level of fatigue. Modifiable factors significantly associated with depression further included a lack of access to wellness resources, lack of time off for leisure, and fewer hours of sleep. Fifty-five percent (41 of 74) of participants reported direct impacts to their personal circumstances by the COVID-19 pandemic. CONCLUSIONS.­: Burnout and depression are significant issues affecting Canadian laboratory medicine residents. As the COVID-19 pandemic continues, we recommend the institution of flexible work arrangements, protected time off for illness and leisure, ongoing evaluation of career satisfaction, formal and informal wellness programming with trainee input, formal mentorship programming, and a financial literacy curriculum as measures to improve trainee wellness.

Quantitative Pathologic Analysis of Digitized Images of Colorectal Carcinoma Improves Prediction of Recurrence-Free Survival.

BACKGROUND & AIMS: To examine whether quantitative pathologic analysis of digitized hematoxylin and eosin slides of colorectal carcinoma (CRC) correlates with clinicopathologic features, molecular alterations, and prognosis. METHODS: A quantitative segmentation algorithm (QuantCRC) was applied to 6468 digitized hematoxylin and eosin slides of CRCs. Fifteen parameters were recorded from each image and tested for associations with clinicopathologic features and molecular alterations. A prognostic model was developed to predict recurrence-free survival using data from the internal cohort (n = 1928) and validated on an internal test (n = 483) and external cohort (n = 938). RESULTS: There were significant differences in QuantCRC according to stage, histologic subtype, grade, venous/lymphatic/perineural invasion, tumor budding, CD8 immunohistochemistry, mismatch repair status, KRAS mutation, BRAF mutation, and CpG methylation. A prognostic model incorporating stage, mismatch repair, and QuantCRC resulted in a Harrell's concordance (c)-index of 0.714 (95% confidence interval [CI], 0.702-0.724) in the internal test and 0.744 (95% CI, 0.741-0.754) in the external cohort. Removing QuantCRC from the model reduced the c-index to 0.679 (95% CI, 0.673-0.694) in the external cohort. Prognostic risk groups were identified, which provided a hazard ratio of 2.24 (95% CI, 1.33-3.87, P = .004) for low vs high-risk stage III CRCs and 2.36 (95% CI, 1.07-5.20, P = .03) for low vs high-risk stage II CRCs, in the external cohort after adjusting for established risk factors. The predicted median 36-month recurrence rate for high-risk stage III CRCs was 32.7% vs 13.4% for low-risk stage III and 15.8% for high-risk stage II vs 5.4% for low-risk stage II CRCs. CONCLUSIONS: QuantCRC provides a powerful adjunct to routine pathologic reporting of CRC. A prognostic model using QuantCRC improves prediction of recurrence-free survival.

A Novel Combined Tumor Budding-Poorly Differentiated Clusters Grading System Predicts Recurrence and Survival in Stage I-III Colorectal Cancer.

Tumor budding (TB) and poorly differentiated clusters (PDCs) are powerful prognostic factors in colorectal cancer (CRC). Despite their morphologic and biological overlap, TB and PDC are assessed separately and are distinguished by an arbitrary cutoff for cell cluster size. This cutoff can be challenging to apply in practice and its biological significance remains unclear. We developed a novel scoring system that incorporates TB and PDC into a single parameter ("Combined Score"; CS), eliminating the need for such cutoffs and allowing the prognostic value of PDC to be captured alongside TB. In a cohort of 481 stage I-III CRC resections, CS was significantly associated with American Joint Committee on Cancer (AJCC) stage, T-stage, N-stage, histologic grade, tumor deposits, lymphovascular invasion, and perineural invasion ( P <0.0001). In addition, CS was significantly associated with decreased 5-year recurrence-free survival, overall survival, and disease-specific survival ( P <0.0001). TB and PDC showed similar associations with oncologic outcomes, with hazard ratios consistently lower than for CS. The association between CS and oncologic outcomes remained significant in subgroup analyses stratified by AJCC stage, anatomic location (rectum/colon) and neoadjuvant therapy status. On multivariable analysis, CS retained its significant association with oncologic outcomes ( P =0.0002, 0.005, and 0.009) for recurrence-free survival, disease-specific survival, and overall survival, respectively. In conclusion, CS provides powerful risk stratification in CRC which is at least equivalent to that of TB and PDC assessed individually. If validated elsewhere, CS has practical advantages and a biological rationale that may make it an attractive alternative to assessing these features separately.

Routine Elastin Staining in Surgically Resected Colorectal Cancer: Impact on Venous Invasion Detection and its Association With Oncologic Outcomes.

Venous invasion (VI) is a powerful yet underreported prognostic factor in colorectal cancer (CRC). Its detection can be improved with an elastin stain. We evaluated the impact of routine elastin staining on VI detection in resected CRC and its relationship with oncologic outcomes. Pathology reports from the year before (n=145) and the year following (n=128) the implementation of routine elastin staining at our institution were reviewed for established prognostic factors, including VI. A second review, using elastin stains, documented the presence/absence, location, number, and size of VI foci. The relationship between VI and oncologic outcomes was evaluated for original and review assessments. VI detection rates increased from 21% to 45% following implementation of routine elastin staining (odds ratio [OR]=3.1; 95% confidence interval [CI]: 1.8-5.3; P<0.0001). The second review revealed a lower VI miss rate postimplementation than preimplementation (22% vs. 48%, respectively; P=0.007); this difference was even greater for extramural VI-positive cases (9% vs. 38%, respectively; P=0.0003). Missed VI cases postimplementation had fewer VI foci per missed case (P=0.02) and a trend towards less extramural VI than those missed preimplementation. VI assessed with an elastin stain was significantly associated with recurrence-free survival (P=0.003), and cancer-specific survival (P=0.01) in contrast to VI assessed on hematoxylin and eosin alone (P=0.053 and 0.1, respectively). The association between VI and hematogenous metastasis was far stronger for elastin-detected VI (OR=11.5; 95% CI: 3.4-37.1; P<0.0001) than for hematoxylin and eosin-detected VI (OR=3.7; 95% CI: 1.4-9.9; P=0.01). Routine elastin staining enhances VI detection and its ability to stratify risk in CRC and should be considered for evaluation of CRC resection specimens.

Feasibility and Performance of Elastin Trichrome as a Primary Stain in Colorectal Cancer Resection Specimens: Results of an Interobserver Variability Study.

Venous invasion (VI) is a powerful prognostic factor in colorectal cancer (CRC) that is widely underreported. The ability of elastin stains to improve VI detection is now recognized in several international CRC pathology protocols. However, concerns related to the cost and time required to perform and evaluate these stains in addition to routine hematoxylin and eosin (H&E) stains remains a barrier to their wider use. We therefore sought to determine whether an elastin trichrome (ET) stain could be used as a "stand-alone" stain in CRC resections, by comparing the sensitivity, accuracy, and reproducibility of detection of CAP-mandated prognostic factors using ET and H&E stains. Representative H&E- and ET-stained slides from 50 CRC resections, including a representative mix of stages and prognostic factors, were used to generate 2 study sets. Each case was represented by H&E slides in 1 study set and by corresponding ET slides from the same blocks in the other study set. Ten observers (3 academic gastrointestinal [GI] pathologists, 4 community pathologists, 3 fellows) evaluated each study set for CAP-mandated prognostic factors. ET outperformed H&E in the assessment of VI with respect to detection rates (50% vs. 28.6%; P<0.0001), accuracy (82% vs. 59%, P<0.0001), and reproducibility (k=0.554 vs. 0.394). No significant differences between ET and H&E were observed for other features evaluated. In a poststudy survey, most observers considered the ease and speed of assessment at least equivalent for ET and H&E for most prognostic factors, and felt that ET would be feasible as a stand-alone stain in practice. If validated by others, our findings support the use of ET, rather than H&E, as the primary stain for the evaluation of CRC resections.

Development and initial validation of a deep learning algorithm to quantify histological features in colorectal carcinoma including tumour budding/poorly differentiated clusters.

AIMS: To develop and validate a deep learning algorithm to quantify a broad spectrum of histological features in colorectal carcinoma. METHODS AND RESULTS: A deep learning algorithm was trained on haematoxylin and eosin-stained slides from tissue microarrays of colorectal carcinomas (N = 230) to segment colorectal carcinoma digitised images into 13 regions and one object. The segmentation algorithm demonstrated moderate to almost perfect agreement with interpretations by gastrointestinal pathologists, and was applied to an independent test cohort of digitised whole slides of colorectal carcinoma (N = 136). The algorithm correctly classified mucinous and high-grade tumours, and identified significant differences between mismatch repair-proficient and mismatch repair-deficient (MMRD) tumours with regard to mucin, inflammatory stroma, and tumour-infiltrating lymphocytes (TILs). A cutoff of >44.4 TILs per mm2 carcinoma gave a sensitivity of 88% and a specificity of 73% in classifying MMRD carcinomas. Algorithm measures of tumour budding (TB) and poorly differentiated clusters (PDCs) outperformed TB grade derived from routine sign-out, and compared favourably with manual counts of TB/PDCs with regard to lymphatic, venous and perineural invasion. Comparable associations were seen between algorithm measures of TB/PDCs and manual counts of TB/PDCs for lymph node metastasis (all P < 0.001); however, stronger correlations were seen between the proportion of positive lymph nodes and algorithm measures of TB/PDCs. Stronger associations were also seen between distant metastasis and algorithm measures of TB/PDCs (P = 0.004) than between distant metastasis and TB (P = 0.04) and TB/PDC counts (P = 0.06). CONCLUSIONS: Our results highlight the potential of deep learning to identify and quantify a broad spectrum of histological features in colorectal carcinoma.

Poorly differentiated clusters in colorectal cancer: a current review and implications for future practice.

Poorly differentiated clusters (PDC), defined as small groups of ≥5 tumour cells without glandular differentiation, have gained recent attention as a promising prognostic factor in colorectal cancer (CRC). Numerous studies have shown PDC to be significantly associated with other adverse histopathological features and worse clinical outcomes. PDC may hold particular promise in stage II colon cancer, where risk stratification plays a critical role in patient selection for adjuvant chemotherapy. In addition, emerging evidence suggests that PDC can predict lymph node metastasis in endoscopically resected pT1 CRC, potentially helping the selection of patients for oncological resection. In 'head-to-head' comparisons, PDC grade has consistently outperformed conventional histological grading systems both in terms of risk stratification and reproducibility. With a number of large-scale studies now available, this review evaluates the evidence regarding the prognostic significance of PDC, considers its relationship with other emerging invasive front prognostic markers (such as tumour budding and stroma type), assesses its 'practice readiness', addressing issues such as interobserver reproducibility, scoring methodologies and special histological subtypes (e.g. micropapillary and mucinous carcinoma), and draws attention to ongoing challenges and areas in need of further study. Finally, emerging data on the role of PDC in non-colorectal cancers are briefly considered.