Grading of oral squamous cell carcinomas - Intra and interrater agreeability: Simpler is better?

Seeing the random forest through the decision trees. Supporting learning health systems from histopathology with machine learning models: Challenges and opportunities

This paper discusses some overlooked challenges faced when working with machine learning models for histopathology and presents a novel opportunity to support "Learning Health Systems" with them. Initially, the authors elaborate on these challenges after separating them according to their mitigation strategies: those that need innovative approaches, time, or future technological capabilities and those that require a conceptual reappraisal from a critical perspective. Then, a novel opportunity to support "Learning Health Systems" by integrating hidden information extracted by ML models from digitalized histopathology slides with other healthcare big data is presented.

The clinical implication and translational research of OSCC differentiation

BACKGROUND

The clinical value and molecular characteristics of tumor differentiation in oral squamous cell carcinoma (OSCC) remain unclear. There is a lack of a related molecular classification prediction system based on pathological images for precision medicine.

METHODS

Integration of epidemiology, genomics, experiments, and deep learning to clarify the clinical value and molecular characteristics, and develop a novel OSCC molecular classification prediction system.

RESULTS

Large-scale epidemiology data (n = 118,817) demonstrated OSCC differentiation was a significant prognosis indicator (p < 0.001), and well-differentiated OSCC was more chemo-resistant than poorly differentiated OSCC. These results were confirmed in the TCGA database and in vitro. Furthermore, we found chemo-resistant related pathways and cell cycle-related pathways were up-regulated in well- and poorly differentiated OSCC, respectively. Based on the characteristics of OSCC differentiation, a molecular grade of OSCC was obtained and combined with pathological images to establish a novel prediction system through deep learning, named ShuffleNetV2-based Molecular Grade of OSCC (SMGO). Importantly, our independent multi-center cohort of OSCC (n = 340) confirmed the high accuracy of SMGO.

CONCLUSIONS

OSCC differentiation was a significant indicator of prognosis and chemotherapy selection. Importantly, SMGO could be an indispensable reference for OSCC differentiation and assist the decision-making of chemotherapy.

Silencing of heat shock factor 1 (HSF1) inhibits proliferation, invasion, and epithelial-mesenchymal transition in oral squamous cell carcinoma

BACKGROUND

Oral squamous cell carcinoma is characterized by high rates of morbidity and mortality. Evidence obtained for different types of cancer shows that tumor initiation, progression, and therapeutic resistance are regulated by heat shock factor 1. This research aimed to analyze the effects of heat shock factor 1 on the biological behavior of oral squamous cell carcinoma.

METHODS

Clinicopathological and immunoexpression study of heat shock factor 1 in 70 cases of oral tongue SCC and functional assays by gene silencing of this factor in an oral tongue SCC cell line.

RESULTS

Heat shock factor 1 was overexpressed in oral tongue SCC specimens compared to normal oral mucosa (p < 0.0001) and in the SCC15 line compared to immortalized keratinocytes (p < 0.005). No significant associations were observed between overexpression of heat shock factor 1 and clinicopathological parameters or survival rates of the oral tongue SCC cases in the present sample. In vitro experiments showed that heat shock factor 1 silencing inhibited cell proliferation (p < 0.005) and cell cycle progression, with the accumulation of cells in the G0/G1 phase (p < 0.01). In addition, heat shock factor 1 silencing reduced cell invasion capacity (p < 0.05) and epithelial-mesenchymal transition, characterized by a decrease in vimentin expression (p < 0.05) and an increase in E-cadherin expression (p < 0.001).

CONCLUSION

Heat shock factor 1 may exert several functions that help maintain cell stability under the stressful conditions of the tumor microenvironment. Thus, strategies targeting the regulation of this protein may in the future be a useful therapeutic tool to control the progression of oral squamous cell carcinoma.

Cellular dissociation grading on biopsies of pulmonary squamous cell carcinoma provides prognostic information across all stages and is congruent with resection specimen grading

Grading of squamous cell carcinomas (SCCs) based on tumour budding and cell nest size has been termed cellular dissociation grading (CDG) and was suggested as a robust outcome predictor when assessed in biopsies and resections of various extrapulmonary SCCs. In pulmonary SCC (pSCC), this has so far been shown only for resected cancers. As most lung cancers are inoperable, it is of utmost importance to clarify whether the prognostic impact of CDG is retained in the biopsy setting. Two independent pSCC biopsy cohorts from Munich (n = 134, non-resected) and Heidelberg (n = 135, resected) were assessed. Tumour budding and cell nest size measures were assembled into the three-tiered CDG system (G1-G3). Data were correlated with clinicopathological parameters and overall- (OS), disease-specific- (DSS), and disease-free survival (DFS). Interobserver variability and concordance between biopsy and resection specimen were also investigated. CDG was highly congruent between biopsy and resection specimens (κ = 0.77, p < 0.001). In both pSCC cohorts, biopsy-derived CDG strongly impacted on OS, DSS, and DFS (e.g. DFS: p < 0.001). In multivariate survival analyses, CDG remained a stage independent predictor of survival in both cohorts (DFS: p < 0.001 respectively; hazard ratio Munich cohort: CDG-G2: 4.31, CDG-G3; 5.14; Heidelberg cohort: CDG-G2: 5.87, CDG-G3: 9.07). Interobserver agreement for CDG was almost perfect (κ = 0.84, p < 0.001). We conclude that assessment of CDG based on tumour budding and cell nest size is feasible on pSCC biopsies and harbours stage independent prognostic information in resectable as well as non-resectable pSCC. Integration of this grading approach into clinicopathological routine should be considered.

Identification of Metabolism-Associated Biomarkers for Early and Precise Diagnosis of Oral Squamous Cell Carcinoma

The 5-year survival rate for oral squamous cell carcinoma (OSCC), one of the most common head and neck cancers, has not improved in the last 20 years. Poor prognosis of OSCC is the result of failure in early and precise diagnosis. Metabolic reprogramming, including the alteration of the uptake and utilisation of glucose, amino acids and lipids, is an important feature of OSCC and can be used to identify its biomarkers for early and precise diagnosis. In this review, we summarise how recent findings of rewired metabolic networks in OSCC have facilitated early and precise diagnosis of OSCC.

Histopathologic grading and its relationship with outcome in oral tongue squamous cell carcinoma

BACKGROUND

Histopathologic grading has been routinely used as a complement for clinical staging in the prognostication of patients with oral tongue squamous cell carcinoma (OTSCC). However, this subject remains contentious because there is no universally accepted grading system.

OBJECTIVES

This study compared the prognostic significance of four histopathologic grading systems in 80 cases of oral tongue squamous cell carcinoma (OTSCC).

METHODS

Clinical and follow-up information of the patients were obtained from medical records. Histopathologic malignancy grading of the tumor invasive front, Histologic risk assessment (HRA), World Health Organization (WHO) grading system, and Budding and Depth of invasion (BD) model were evaluated in the surgical specimens.

RESULTS

The HRA, histopathologic malignancy grading and WHO systems did not predict survival. Patients with larger tumor size [Hazard ratio (HR): 2.38; 95% confidence interval (CI): 1.07-5.27; P = 0.026] and patients with BD model high-grade tumors (HR: 2.99; 95% CI: 1.03-8.68; P = 0.034) were significantly associated with a poor 5-year overall survival rate. In the multivariate analysis, tumor size was identified as the only significant independent prognostic factor (HR: 2.23; 95% CI: 1.00-4.99; P = 0.050). None of the grading systems studied was associated with 5-year disease-free survival rates.

CONCLUSIONS

BD model was the only histopathologic grading system associated with the outcome of patients with OTSCC, indicating its potential value as an effective tool for the prognostication of OTSCC.

Survival and clinicopathological characteristics of different histological grades of oral cavity squamous cell carcinoma: A single-center retrospective study

The TNM staging system for oral squamous cell carcinoma (OSCC) provides clinicians a dependable foundation for patient prognosis and management decisions, but in clinical practice, treatment outcomes of patients with OSCC are sometimes unsatisfactory. This retrospective study investigated the association between survival and clinicopathological characteristics and histological grades of 2535 patients with OSCC. Additionally, the present study aimed to compare the predictive abilities of histological grades with other common prognostic factors. The enrolled patients were divided into three groups by two experienced pathologists into well-differentiated, moderately differentiated, and poorly differentiated groups, according to the WHO classification. Finally, we designed an observational, retrospective study based on the histological grading of tumors to compare their clinicopathological characteristics and conducted survival analysis among the three groups. Advanced tumor stage was diagnosed in 23.9%, 44.0%, and 55.1% of patients with grades 1–3 OSCC, respectively. By T status, T3 or T4 tumors were found in approximately 22%, 34%, and 40% of patients with grades 1–3 OSCC, respectively. By N status, lymph node metastases were found in 6.1%, 29.3%, and 45.9% of patients with grades 1–3 OSCC, respectively. Thus, significant survival differences were observed based on different OSCC histological grades. Meanwhile, in the multivariate (adjusted) analysis, N1 and N2 stages, extranodal spread, and poor differentiation were associated with a higher recurrence risk than the other common prognostic factors. In conclusion, 5% of patients in our study presented with poorly differentiated OSCC at diagnosis. Furthermore, grade 3 OSCC has worse prognosis and is more aggressive than grades 1 and 2 OSCC. In the future, we should focus on modifying individual therapy for poorly differentiated OSCC to achieve improved outcomes.

A combined histo-score based on tumor differentiation and lymphocytic infiltrate is a robust prognostic marker for mobile tongue cancer

We wanted to evaluate the prognostic value of common histopathological variables in a large cohort of patients with cancer in the mobile tongue as such information can be important for treatment stratification of the individual patient, especially for patients with low-stage disease. In addition, we wanted to investigate whether an alternative scoring model with fewer options would compromise the prognostic value. One hundred fifty patients with oral tongue squamous cell carcinomas that were treated in curative intent and with available HE-stained tumor sections were included. We reclassified all tumors and performed univariate and multivariate survival analyses of histopathological and clinical variables. For the complete cohort, lymph node status, grade of differentiation, perineural infiltration, and lymphocytic infiltration were independent prognosticators. In the low-stage disease group, independent prognostic factors were tumor size, grade of differentiation, and lymphocytic infiltrate. For patients with low-stage disease, a histo-score combining the scores for tumor differentiation and lymphocytic infiltrate identified a group of patients with particularly low survival, as patients with moderately or poorly differentiated tumors and little lymphocytic infiltrate had a less favorable 5-year survival outcome than patients in the high-stage disease group. This study shows that a histo-score combining tumor differentiation and lymphocytic infiltration should be given special consideration in treatment planning. Our results also illustrate that many variables can be scored with fewer options than previously suggested to increase their reproducibility, and still maintain their prognostic value.

Staging more important than grading? Evaluation of malignancy grading, depth of invasion, and resection margins in oral squamous cell carcinoma

Objectives

The present study evaluated the predictive value of staging and grading parameters concerning the presence of lymph-node metastases, overall survival (OS), and relapse-free survival (RFS) of patients with oral squamous cell carcinoma (OSCC).

Materials and methods

HE-stains of 135 surgically treated (R0) primary OSCCs were analyzed using a both microscopic and software-based approach. Depth of invasion (DOI) and resection margins (RM) were measured, and each case was graded according to the malignancy grading system as described by Anneroth et al. and Bryne et al. on two different sites of the tumor (surface and invasion front; TS and IF).

Results

Parameters that could be identified as significant predictors of OS and RFS were UICC cancer stage (p = 0.009 and p = 0.012); pT-stage as defined in the 7th edition (p = 0.029 and 0.015) and, after restaging using DOI, 8th edition (p = 0.023 and p = 0.005) of the TNM classification of malignant tumors; the presence of lymphonodular metastases (LM) (p = 0.004 and p = 0.011); degree of keratinization (p = 0.029 and p = 0.042); and pattern of growth (p = 0.029 and p = 0.024) at the TS after applying a binary scale for both parameters. Also, when directly comparing the most extreme subgroups (scores 1 and 4) of lymphoplasmacytic infiltration at the IF, there was a significant difference in OS (p = 0.046) and RFS (p = 0.005). Invasion of blood vessels (p = 0.013) and perineural invasion (p = 0.023) were significantly associated with a lower OS. Age lower than 60 years (univariate p = 0.029, multivariate p = 0.031), infiltration of lymphatic vessels (p = 0.003), infiltration of nerves (p = 0.010), pT-stage (8th edition) (p = 0.014), degree of keratinization at the IF (p = 0.033), and nuclear polymorphism at the IF (p = 0.043) after conversion to a binary scale were found to be significant prognostic parameters regarding the presence of LM. DOI evolved as a significant predictor for OS (p = 0.006), RFS (p = 0.003), and LM (p = 0.032) in metric and grouped analysis.

Conclusions

The current evaluation revealed depth of invasion as strongest histologic predictor of metastatic tumor growth, overall survival, and relapse-free survival in OSCC, confirming the current adaption of the T-classification. Other distinct histologic grading parameters investigated during this study can give valuable indications of a tumor’s potential aggressiveness, but the exact site, mode, and procedure need further exploration.

Clinical relevance

Integrating measurement of DOI also into the pretherapeutic staging process could aid in treatment planning.