Radiomics diagnosed histopathological growth pattern in prediction of response and 1-year progression free survival for colorectal liver metastases patients treated with bevacizumab containing chemotherapy

Leveraging radiomics and AI for precision diagnosis and prognostication of liver malignancies

Liver tumors, whether primary or metastatic, have emerged as a growing concern with substantial global health implications. Timely identification and characterization of liver tumors are pivotal factors in order to provide optimum treatment. Imaging is a crucial part of the detection of liver tumors; however, conventional imaging has shortcomings in the proper characterization of these tumors which leads to the need for tissue biopsy. Artificial intelligence (AI) and radiomics have recently emerged as investigational opportunities with the potential to enhance the detection and characterization of liver lesions. These advancements offer opportunities for better diagnostic accuracy, prognostication, and thereby improving patient care. In particular, these techniques have the potential to predict the histopathology, genotype, and immunophenotype of tumors based on imaging data, hence providing guidance for personalized treatment of such tumors. In this review, we outline the progression and potential of AI in the field of liver oncology imaging, specifically emphasizing manual radiomic techniques and deep learning-based representations. We discuss how these tools can aid in clinical decision-making challenges. These challenges encompass a broad range of tasks, from prognosticating patient outcomes, differentiating benign treatment-related factors and actual disease progression, recognizing uncommon response patterns, and even predicting the genetic and molecular characteristics of the tumors. Lastly, we discuss the pitfalls, technical limitations and future direction of these AI-based techniques.

Prediction of transformation in the histopathological growth pattern of colorectal liver metastases after chemotherapy using CT-based radiomics

Chemotherapy alters the prognostic biomarker histopathological growth pattern (HGP) phenotype in colorectal liver metastases (CRLMs) patients. We aimed to develop a CT-based radiomics model to predict the transformation of the HGP phenotype after chemotherapy. This study included 181 patients with 298 CRLMs who underwent preoperative contrast-enhanced CT followed by partial hepatectomy between January 2007 and July 2022 at two institutions. HGPs were categorized as pure desmoplastic HGP (pdHGP) or non-pdHGP. The samples were allocated to training, internal validation, and external validation cohorts comprising 153, 65, and 29 CRLMs, respectively. Radiomics analysis was performed on pre-enhanced, arterial phase, portal venous phase (PVP), and fused images. The model was used to predict prechemotherapy HGPs in 112 CRLMs, and HGP transformation was analysed by comparing these findings with postchemotherapy HGPs determined pathologically. The prevalence of pdHGP was 19.8% (23/116) and 45.8% (70/153) in chemonaïve and postchemotherapy patients, respectively (P < 0.001). The PVP radiomics signature showed good performance in distinguishing pdHGP from non-pdHGPs (AUCs of 0.906, 0.877, and 0.805 in the training, internal validation, and external validation cohorts, respectively). The prevalence of prechemotherapy pdHGP predicted by the radiomics model was 33.0% (37/112), and the prevalence of postchemotherapy pdHGP according to the pathological analysis was 47.3% (53/112; P = 0.029). The transformation of HGP was bidirectional, with 15.2% (17/112) of CRLMs transforming from prechemotherapy pdHGP to postchemotherapy non-pdHGP and 30.4% (34/112) transforming from prechemotherapy non-pdHGP to postchemotherapy pdHGP (P = 0.005). CT-based radiomics method can be used to effectively predict the HGP transformation in chemotherapy-treated CRLM patients, thereby providing a basis for treatment decisions.

Tumor-liver interface in MRI of liver metastasis enables prediction of EGFR mutation in patients with lung cancer: A proof-of-concept study

BACKGROUND

Preoperative prediction of the epidermal growth factor receptor (EGFR) status in non-small-cell lung cancer (NSCLC) patients with liver metastasis (LM) may have potential clinical values for assisting in treatment decision-making.

PURPOSE

To explore the value of tumor-liver interface (TLI)-based magnetic resonance imaging (MRI) radiomics for detecting the EGFR mutation in NSCLC patients with LM.

METHODS

This retrospective study included 123 and 44 patients from hospital 1 (between Feb. 2018 and Dec. 2021) and hospital 2 (between Nov. 2015 and Aug. 2022), respectively. The patients received contrast-enhanced T1-weighted (CET1) and T2-weighted (T2W) liver MRI scans before treatment. Radiomics features were extracted from MRI images of TLI and the whole tumor region, separately. The least absolute shrinkage and selection operator (LASSO) regression was used to screen the features and establish radiomics signatures (RSs) based on TLI (RS-TLI) and the whole tumor (RS-W). The RSs were evaluated by the receiver operating characteristic (ROC) curve analysis.

RESULTS

A total of 5 and 6 features were identified highly correlated with the EGFR mutation status from TLI and the whole tumor, respectively. The RS-TLI showed better prediction performance than RS-W in the training (AUCs, RS-TLI vs. RS-W, 0.842 vs. 0.797), internal validation (AUCs, RS-TLI vs. RS-W, 0.771 vs. 0.676) and external validation (AUCs, RS-TLI vs. RS-W, 0.733 vs. 0.679) cohort.

CONCLUSION

Our study demonstrated that TLI-based radiomics can improve prediction performance of the EGFR mutation in lung cancer patients with LM. The established multi-parametric MRI radiomics models may be used as new markers that can potentially assist in personalized treatment planning.

CT-based radiomics analysis to predict local progression of recurrent colorectal liver metastases after microwave ablation

The objective of this study is to establish and validate a radiomics nomogram for prediction of local tumor progression (LTP) after microwave ablation (MWA) for recurrent colorectal liver metastases (CRLM) after hepatic resection. We included 318 consecutive recurrent CRLM patients (216 of training while 102 of validation cohort) with contrast-enhanced computerized tomography images treated with MWA between January 2014 and October 2018. Support vector machine-generated radiomics signature was incorporated together with clinical information to establish a radiomics nomogram. Our constructed radiomics signature including 15 features (first-order intensity statistics features, shape and size-based features, gray level size zone/dependence matrix features) performed well in assessing LTP for both cohorts. With regard to its predictive performance, its C-index was 0.912, compared to the clinical or radiomics models only (c-statistic 0.89 and 0.75, respectively) in the training cohort. In the validation cohort, the radiomics nomogram had better performance (area under the curve = 0.89) compared to the radiomics and clinical models (0.85 and 0.69). According to decision curve analysis, our as-constructed radiomics nomogram showed high clinical utility. As revealed by survival analysis, LTP showed worse progression-free survival (3-year progression-free survival 42.6% vs 78.4%, P < .01). High-risk patients identified using this radiomics signature exhibited worse LTP compared with low-risk patients (3-year LTP 80.2% vs 48.6%, P < .01). A radiomics-based nomogram of pre-ablation computerized tomography imaging may be the precious biomarker model for predicting LTP and personalized risk stratification for recurrent CRLM after hepatic resection treated by MWA.

Histopathological growth patterns and tumor-infiltrating lymphocytes in breast cancer liver metastases

Liver is the third most common organ for breast cancer (BC) metastasis. Two main histopathological growth patterns (HGP) exist in liver metastases (LM): desmoplastic and replacement. Although a reduced immunotherapy efficacy is reported in patients with LM, tumor-infiltrating lymphocytes (TIL) have not yet been investigated in BCLM. Here, we evaluate the distribution of the HGP and TIL in BCLM, and their association with clinicopathological variables and survival. We collect samples from surgically resected BCLM (n = 133 patients, 568 H&E sections) and post-mortem derived BCLM (n = 23 patients, 97 H&E sections). HGP is assessed as the proportion of tumor liver interface and categorized as pure-replacement ('pure r-HGP') or any-desmoplastic ('any d-HGP'). We score the TIL according to LM-specific guidelines. Associations with progression-free (PFS) and overall survival (OS) are assessed using Cox regressions. We observe a higher prevalence of 'any d-HGP' (56%) in the surgical samples and a higher prevalence of 'pure r-HGP' (83%) in the post-mortem samples. In the surgical cohort, no evidence of the association between HGP and clinicopathological characteristics is observed except with the laterality of the primary tumor (p value = 0.049) and the systemic preoperative treatment before liver surgery (p value = .039). TIL is less prevalent in 'pure r-HGP' as compared to 'any d-HGP' (p value = 0.001). 'Pure r-HGP' predicts worse PFS (HR: 2.65; CI: (1.45-4.82); p value = 0.001) and OS (HR: 3.10; CI: (1.29-7.46); p value = 0.011) in the multivariable analyses. To conclude, we demonstrate that BCLM with a 'pure r-HGP' is associated with less TIL and with the worse outcome when compared with BCLM with 'any d-HGP'. These findings suggest that HGP could be considered to refine treatment approaches.

Radiomics and machine learning analysis by computed tomography and magnetic resonance imaging in colorectal liver metastases prognostic assessment

OBJECTIVE

The aim of this study was the evaluation radiomics analysis efficacy performed using computed tomography (CT) and magnetic resonance imaging in the prediction of colorectal liver metastases patterns linked to patient prognosis: tumor growth front; grade; tumor budding; mucinous type. Moreover, the prediction of liver recurrence was also evaluated.

METHODS

The retrospective study included an internal and validation dataset; the first was composed by 119 liver metastases from 49 patients while the second consisted to 28 patients with single lesion. Radiomic features were extracted using PyRadiomics. Univariate and multivariate approaches including machine learning algorithms were employed.

RESULTS

The best predictor to identify tumor growth was the Wavelet_HLH_glcm_MaximumProbability with an accuracy of 84% and to detect recurrence the best predictor was wavelet_HLH_ngtdm_Complexity with an accuracy of 90%, both extracted by T1-weigthed arterial phase sequence. The best predictor to detect tumor budding was the wavelet_LLH_glcm_Imc1 with an accuracy of 88% and to identify mucinous type was wavelet_LLH_glcm_JointEntropy with an accuracy of 92%, both calculated on T2-weigthed sequence. An increase statistically significant of accuracy (90%) was obtained using a linear weighted combination of 15 predictors extracted by T2-weigthed images to detect tumor front growth. An increase statistically significant of accuracy at 93% was obtained using a linear weighted combination of 11 predictors by the T1-weigthed arterial phase sequence to classify tumor budding. An increase statistically significant of accuracy at 97% was obtained using a linear weighted combination of 16 predictors extracted on CT to detect recurrence. An increase statistically significant of accuracy was obtained in the tumor budding identification considering a K-nearest neighbors and the 11 significant features extracted T1-weigthed arterial phase sequence.

CONCLUSIONS

The results confirmed the Radiomics capacity to recognize clinical and histopathological prognostic features that should influence the choice of treatments in colorectal liver metastases patients to obtain a more personalized therapy.

Deep radiomics-based fusion model for prediction of bevacizumab treatment response and outcome in patients with colorectal cancer liver metastases: a multicentre cohort study

Background

Accurate tumour response prediction to targeted therapy allows for personalised conversion therapy for patients with unresectable colorectal cancer liver metastases (CRLM). In this study, we aimed to develop and validate a multi-modal deep learning model to predict the efficacy of bevacizumab in patients with initially unresectable CRLM using baseline PET/CT, clinical data, and colonoscopy biopsy specimens.

Methods

In this multicentre cohort study, we retrospectively collected data of 307 patients with CRLM from the BECOME study (NCT01972490) (Zhongshan Hospital of Fudan University, Shanghai) and two independent Chinese cohorts (internal validation cohort from January 1, 2018 to December 31, 2018 at Zhongshan Hospital of Fudan University; external validation cohort from January 1, 2020 to December 31, 2020 at Zhongshan Hospital-Xiamen, Shanghai, and the First Hospital of Wenzhou Medical University, Wenzhou). The main inclusion criteria were that patients with CRLM had pre-treatment PET/CT images as well as colonoscopy specimens. After extracting PET/CT features with deep neural networks (DNN) and selecting related clinical factors using LASSO analysis, a random forest classifier was built as the Deep Radiomics Bevacizumab efficacy predicting model (DERBY). Furthermore, by combining histopathological biomarkers into DERBY, we established DERBY+. The performance of model was evaluated using area under the curve (AUC), sensitivity, specificity, positive predictive value, and negative predictive value.

Findings

DERBY achieved promising performance in predicting bevacizumab sensitivity with an AUC of 0.77 and 95% confidence interval (CI) [0.67-0.87]. After combining histopathological features, we developed DERBY+, which had more robust accuracy for predicting tumour response in external validation cohort (AUC 0.83 and 95% CI [0.75-0.92], sensitivity 80.4%, specificity 76.8%). DERBY+ also had prognostic value: the responders had longer progression-free survival (median progression-free survival: 9.6 vs 6.3 months, p = 0.002) and overall survival (median overall survival: 27.6 vs 18.5 months, p = 0.010) than non-responders.

Interpretation

This multi-modal deep radiomics model, using PET/CT, clinical data and histopathological data, was able to identify patients with bevacizumab-sensitive CRLM, providing a favourable approach for precise patient treatment. To further validate and explore the clinical impact of this work, future prospective studies with larger patient cohorts are warranted.

Funding

The National Natural Science Foundation of China; Fujian Provincial Health Commission Project; Xiamen Science and Technology Agency Program; Clinical Research Plan of SHDC; Shanghai Science and Technology Committee Project; Clinical Research Plan of SHDC; Zhejiang Provincial Natural Science Foundation of China; and National Science Foundation of Xiamen.

Histopathological growth pattern of liver metastases as an independent marker of metastatic behavior in different primary cancers

Surgical resection can lead to prolonged survival in patients with isolated liver metastases (LM) from various primary cancers. However, there are currently no validated predictive markers to discriminate between these oligo/argometastatic patients, who will benefit from surgery, and those with diffuse metastatic behavior in whom surgery will be futile. To evaluate whether the tumor microenvironment, or histopathological growth pattern (HGP), of LM reflects the type of metastatic progression independently of the origin of the primary cancer, we analyzed a combined series of patients who underwent surgery for colorectal LM (N=263) or non-colorectal LM (N=66). HGPs of LM were scored in each patient to distinguish between desmoplastic HGP (all LM showing a complete encapsulated pattern) and non-desmoplastic HGP (at least one LM with some infiltrating-replacement component). In the entire series, 5-year overall and progression-free survival were, 44.5% and 15.5%, respectively, with no significant differences between colorectal and non-colorectal LM. In patients with desmoplastic HGP, 5-year overall and progression-free survival were 57% and 32%, respectively, as compared to 41% and 12%, respectively, in patients with non-desmoplastic-HGP (p=0.03 and 0.005). Irrespective of cancer origin and compared to traditional risk factors, desmoplastic HGP was the most significant predictor for better post-operative overall survival (adjusted HR: 0.62; 95% CI: [0.49-0.97]; p=0.035) and progression-free survival (adjusted HR: 0.61; 95% CI: [0.42-0.87], p=0.006). This suggests that the HGP of LM may represent an accurate marker that reflects the mode of metastatic behavior, independently of primary cancer type.

Colorectal liver metastases patients prognostic assessment: prospects and limits of radiomics and radiogenomics

In this narrative review, we reported un up-to-date on the role of radiomics to assess prognostic features, which can impact on the liver metastases patient treatment choice. In the liver metastases patients, the possibility to assess mutational status (RAS or MSI), the tumor growth pattern and the histological subtype (NOS or mucinous) allows a better treatment selection to avoid unnecessary therapies. However, today, the detection of these features require an invasive approach. Recently, radiomics analysis application has improved rapidly, with a consequent growing interest in the oncological field. Radiomics analysis allows the textural characteristics assessment, which are correlated to biological data. This approach is captivating since it should allow to extract biological data from the radiological images, without invasive approach, so that to reduce costs and time, avoiding any risk for the patients. Several studies showed the ability of Radiomics to identify mutational status, tumor growth pattern and histological type in colorectal liver metastases. Although, radiomics analysis in a non-invasive and repeatable way, however features as the poor standardization and generalization of clinical studies results limit the translation of this analysis into clinical practice. Clear limits are data-quality control, reproducibility, repeatability, generalizability of results, and issues related to model overfitting.

Machine Learning Approaches with Textural Features to Calculate Breast Density on Mammography

BACKGROUND

breast cancer (BC) is the world's most prevalent cancer in the female population, with 2.3 million new cases diagnosed worldwide in 2020. The great efforts made to set screening campaigns, early detection programs, and increasingly targeted treatments led to significant improvement in patients' survival. The Full-Field Digital Mammograph (FFDM) is considered the gold standard method for the early diagnosis of BC. From several previous studies, it has emerged that breast density (BD) is a risk factor in the development of BC, affecting the periodicity of screening plans present today at an international level.

OBJECTIVE

in this study, the focus is the development of mammographic image processing techniques that allow the extraction of indicators derived from textural patterns of the mammary parenchyma indicative of BD risk factors.

METHODS

a total of 168 patients were enrolled in the internal training and test set while a total of 51 patients were enrolled to compose the external validation cohort. Different Machine Learning (ML) techniques have been employed to classify breasts based on the values of the tissue density. Textural features were extracted only from breast parenchyma with which to train classifiers, thanks to the aid of ML algorithms.

RESULTS

the accuracy of different tested classifiers varied between 74.15% and 93.55%. The best results were reached by a Support Vector Machine (accuracy of 93.55% and a percentage of true positives and negatives equal to TPP = 94.44% and TNP = 92.31%). The best accuracy was not influenced by the choice of the features selection approach. Considering the external validation cohort, the SVM, as the best classifier with the 7 features selected by a wrapper method, showed an accuracy of 0.95, a sensitivity of 0.96, and a specificity of 0.90.

CONCLUSIONS

our preliminary results showed that the Radiomics analysis and ML approach allow us to objectively identify BD.

L1CAM and laminin vascular network: Association with the high-risk replacement histopathologic growth pattern in uveal melanoma liver metastases

The replacement histopathologic growth pattern (rHGP) in melanoma liver metastases connotes an aggressive phenotype (vascular co-option; angiotropic extravascular migratory spread) and adverse prognosis. Herein, replacement and desmoplastic HGP (dHGP) were studied in uveal melanoma liver metastases (MUM). In particular, L1CAM and a "laminin vascular network" were detected at the advancing front of 14/20 cases (p = 0.014) and 16/20 cases (p = 6.4e-05) rHGPs, respectively, but both were absent in the dHGP (8/8 cases) (p = 0.014, and p = 6.3e-05, respectively). L1CAM highlighted progressive extension of angiotropic melanoma cells along sinusoidal vessels in a pericytic location (pericytic mimicry) into the hepatic parenchyma. An inverse relationship between L1CAM expression and melanin index (p = 0.012) suggested differentiation toward an amelanotic embryonic migratory phenotype in rHGP. Laminin labeled the basement membrane zone interposed between sinusoidal vascular channels and angiotropic melanoma cells at the advancing front. Other new findings: any percentage of rHGP and pure rHGP had a significant adverse effect on metastasis-specific overall survival (p = 0.038; p = 0.0064), as well as predominant rHGP (p = 0.0058). Pure rHGP also was associated with diminished metastasis-free survival relative to dHGP (p = 0.040), possibly having important implications for mechanisms of tumor spread. In conclusion, we report for the first time that L1CAM and a laminin vascular network are directly involved in this high-risk replacement phenotype. Further, this study provides more detailed information about the adverse prognostic effect of the rHGP in MUM.

Histopathological growth patterns of liver metastasis: updated consensus guidelines for pattern scoring, perspectives and recent mechanistic insights

The first consensus guidelines for scoring the histopathological growth patterns (HGPs) of liver metastases were established in 2017. Since then, numerous studies have applied these guidelines, have further substantiated the potential clinical value of the HGPs in patients with liver metastases from various tumour types and are starting to shed light on the biology of the distinct HGPs. In the present guidelines, we give an overview of these studies, discuss novel strategies for predicting the HGPs of liver metastases, such as deep-learning algorithms for whole-slide histopathology images and medical imaging, and highlight liver metastasis animal models that exhibit features of the different HGPs. Based on a pooled analysis of large cohorts of patients with liver-metastatic colorectal cancer, we propose a new cut-off to categorise patients according to the HGPs. An up-to-date standard method for HGP assessment within liver metastases is also presented with the aim of incorporating HGPs into the decision-making processes surrounding the treatment of patients with liver-metastatic cancer. Finally, we propose hypotheses on the cellular and molecular mechanisms that drive the biology of the different HGPs, opening some exciting preclinical and clinical research perspectives.

CT, MRI, and radiomics studies of liver metastasis histopathological growth patterns: an up-to-date review

The histopathological growth patterns (HGPs) of liver metastases (LMs) are independently associated with the long-term prognosis of the primary tumor, with different HGPs predicting different patient outcomes and clinical treatment decisions. Non-invasive imaging biomarkers for stratification of HGPs are beneficial for treatment monitoring, evaluation of efficacy, and prognosis prediction of LMs. This review describes the state of research regarding computed tomography (CT), magnetic resonance imaging (MRI), and radiomics imaging biomarkers for LM-HGPs; discusses the advantages of CT, MRI, and radiomics for classification of LM-HGPs; and provides a reference for the stratification of LM-HGPs. Finally, the difficulties and deficiencies of CT, MRI, and radiomics in LM-HGP research are summarized along with the proposed directions for future research.

Clinical implications and mechanism of histopathological growth pattern in colorectal cancer liver metastases

Liver is the most common site of metastases of colorectal cancer, and liver metastases present with distinct histopathological growth patterns (HGPs), including desmoplastic, pushing and replacement HGPs and two rare HGPs. HGP is a miniature of tumor-host reaction and reflects tumor biology and pathological features as well as host immune dynamics. Many studies have revealed the association of HGPs with carcinogenesis, angiogenesis, and clinical outcomes and indicates HGP functions as bond between microscopic characteristics and clinical implications. These findings make HGP a candidate marker in risk stratification and guiding treatment decision-making, and a target of imaging observation for patient screening. Of note, it is crucial to determine the underlying mechanism shaping HGP, for instance, immune infiltration and extracellular matrix remodeling in desmoplastic HGP, and aggressive characteristics and special vascularization in replacement HGP (rHGP). We highlight the importance of aggressive features, vascularization, host immune and organ structure in formation of HGP, hence propose a novel "advance under camouflage" hypothesis to explain the formation of rHGP.

Histopathological factors versus margin size in single colorectal liver metastases: Does a 1-cm margin size matter?

Background and objective:

The ideal margin width for surgical resection of colorectal liver metastases has been extensively studied, but not sufficiently in accordance with other pathological factors. The aim of this study was to assess for the first time the prognostic impact of margin widths according to different prognostic pathological factors in colorectal liver metastasis.

Methods:

We evaluated 101 patients with a single resected metastasis. Slides stained by HE were assessed for the presence of poorly differentiated clusters, peritumoral inflammatory infiltrate, tumor pseudocapsule, and tumor borders pattern. Overall survival, disease-free survival, and hepatic recurrence were evaluated. The pathologic factors prognostic impact was evaluated according to a (< or ⩾) 10-mm margin size.

Results:

Factors independently associated with a shorter overall survival were absence of tumor pseudocapsule ( p < 0.001) and infiltrative tumor border pattern ( p = 0.019). The absence of tumor pseudocapsule was the only factor independently associated with shorter disease-free survival ( p < 0.001). Hepatic recurrence was associated with infiltrative tumor border and absence of pseudocapsule. Margin width ⩾10 mm did not impact overall survival independently of the studied histological prognostic factors.

Conclusions:

In colorectal liver metastasis resection, the absence of tumor pseudocapsule was significantly associated with shorter overall survival and disease-free survival and hepatic recurrence. However, margins larger than 10 mm did not offer survival benefit when other pathologic negative prognostic factors were concomitantly analyzed, reinforcing the idea that biology, rather than margin size, is crucial for prognosis.