Figure 5 - uploaded by Javier Torres-Roca
Content may be subject to copyright.
Radiosensitivity index distinguishes clinical populations with different disease-related outcomes in head and neck cancer
Source publication
The last two decades have seen technological developments that have led to more accurate delivery of radiation therapy (RT), which has resulted in clinical gains in many solid tumors. However, a fundamental question and perhaps the next major hurdle is whether biological strategies can be developed to further enhance the effectiveness and efficienc...
Citations
... Radiotherapy remains a mainstay of cancer treatment, with approximately half of all cancer patients receiving radiotherapy as part of their standard of care (Fowler 2006;Torres-Roca 2012;Enderling et al. 2009). It is common for a patient's course of treatment to be determined solely by tumour etiology, location, and stage. ...
Longitudinal tumour volume data from head-and-neck cancer patients show that tumours of comparable pre-treatment size and stage may respond very differently to the same radiotherapy fractionation protocol. Mathematical models are often proposed to predict treatment outcome in this context, and have the potential to guide clinical decision-making and inform personalised fractionation protocols. Hindering effective use of models in this context is the sparsity of clinical measurements juxtaposed with the model complexity required to produce the full range of possible patient responses. In this work, we present a compartment model of tumour volume and tumour composition, which, despite relative simplicity, is capable of producing a wide range of patient responses. We then develop novel statistical methodology and leverage a cohort of existing clinical data to produce a predictive model of both tumour volume progression and the associated level of uncertainty that evolves throughout a patient’s course of treatment. To capture inter-patient variability, all model parameters are patient specific, with a bootstrap particle filter-like Bayesian approach developed to model a set of training data as prior knowledge. We validate our approach against a subset of unseen data, and demonstrate both the predictive ability of our trained model and its limitations.
Supplementary Information
The online version contains supplementary material available at 10.1007/s11538-023-01246-0.
... Recent efforts include more strategic integration of basic science approaches into radiobiology and radiation oncology to help better understand the mechanisms of radiation response dynamics and to help predict how to best personalize radiation to individual patients. Genomic signatures (2)(3)(4)(5), imaging metrics (6)(7)(8), and burgeoning machine learning and artificial intelligence approaches are being retrospectively and prospectively evaluated as novel biomarkers for radiation response (9)(10)(11). ...
Introduction
Radiation therapy (RT) is one of the most common anticancer therapies. Yet, current radiation oncology practice does not adapt RT dose for individual patients, despite wide interpatient variability in radiosensitivity and accompanying treatment response. We have previously shown that mechanistic mathematical modeling of tumor volume dynamics can simulate volumetric response to RT for individual patients and estimation personalized RT dose for optimal tumor volume reduction. However, understanding the implications of the choice of the underlying RT response model is critical when calculating personalized RT dose.
Methods
In this study, we evaluate the mathematical implications and biological effects of 2 models of RT response on dose personalization: (1) cytotoxicity to cancer cells that lead to direct tumor volume reduction (DVR) and (2) radiation responses to the tumor microenvironment that lead to tumor carrying capacity reduction (CCR) and subsequent tumor shrinkage. Tumor growth was simulated as logistic growth with pre-treatment dynamics being described in the proliferation saturation index (PSI). The effect of RT was simulated according to each respective model for a standard schedule of fractionated RT with 2 Gy weekday fractions. Parameter sweeps were evaluated for the intrinsic tumor growth rate and the radiosensitivity parameter for both models to observe the qualitative impact of each model parameter. We then calculated the minimum RT dose required for locoregional tumor control (LRC) across all combinations of the full range of radiosensitvity and proliferation saturation values.
Results
Both models estimate that patients with higher radiosensitivity will require a lower RT dose to achieve LRC. However, the two models make opposite estimates on the impact of PSI on the minimum RT dose for LRC: the DVR model estimates that tumors with higher PSI values will require a higher RT dose to achieve LRC, while the CCR model estimates that higher PSI values will require a lower RT dose to achieve LRC.
Discussion
Ultimately, these results show the importance of understanding which model best describes tumor growth and treatment response in a particular setting, before using any such model to make estimates for personalized treatment recommendations.
... Further, three novel genes, RBAP48, RGS19 and TOP1 identified as radiation response predictors for a personalized clinical target of radio sensitization. [135] A lex a nder et al., (2017 ) ident i f ied c yclin E overexpression as a biomarker for combined therapy in inflammatory breast cancer (IBC). Higher expression of cyclin E suggests an aggressiveness and poor prognosis. ...
... The clinical approach of radiosensitivity index (RSI) enhances to complete pathological response by delivering radiation doses to patients predicted as radiosensitive. [135] ATM, BARD1, BRCA1, BRCA2, FANCA, MDC1, MSH2, RAD51 ...
Breast cancer mortality rate is fifth among all cancer and increasing day by day due to modern lifestyles. Its molecular subtype is classified as per their significant receptor expression, such as estrogen receptor (ER), progesterone receptor (PR) & human epidermal growth receptor 2 (Her2). Triple-negative breast cancer (TNBC) is an aggressive subgroup among breast cancer subtypes and clinically challenging to treat due to loss of all three receptor (ER/PR/Her2) expression. Treatment modalities of TNBC include surgery, chemotherapy, radiotherapy and immunotherapy. Postoperative radiation therapy (RT) improves locoregional control and overall survival in TNBC patients. The powerful ionizing radiation (IR) response to RT is contributed by the inherent radiosensitivity of the tumor, which is influenced by genes associated with the cell cycle, DNA damage repair, apoptosis, etc. This review article narrates the role of biomarkers obtained through data mining and manual curation of published literature to predict radioresistance in patients receiving radiotherapy. Further, the role of natural radiosensitizers in overcoming radioresistance for effectively managing TNBC is also discussed.
... Torres-Roca et al. proposed a radiation sensitivity index (RSI), in order to modulate treatment for a subgroup of 40-45% of rectal cancer cases considered not respondent to neo-adjuvant chemo-radiotherapy. The authors also noted a 20% complete pathological response, considering the critical role of radio-sensitivity in the tumor control of locally advanced rectal adenocarcinoma [18]. Extending the applicability of the RSI to head and neck cancers, the authors propose the association of RSI with known prognostic factors including HPV status in order to identify cases that would benefit from an intensification of treatment or radio-sensitizing agents. ...
... RSI is based on a ten-gene network, trained and perfected on 48 HNSCC cell lines and subsequently validated on 5 patient lots including 621 cases. Even if it is not yet considered translatable into clinical practice, RSI opens new horizons in molecular-guided radiotherapy based on radio-sensitivity criteria [15][16][17][18]. ...
Altered fractionation concepts and especially moderate hypo-fractionation are evaluated as alternatives to standard treatment for head and neck squamous cell carcinoma (HNSCC), associated with or not concurrent with or sequential to chemotherapy. The calculation of the iso-equivalent dose regimens has as its starting point the linear quadratic (LQ) formalism traditionally based on the "4Rs" of radiobiology. The higher rates of therapeutic failure after radiotherapy of HNSCC are associated with the heterogeneity of radio-sensibility. The identification of genetic signatures and radio-resistance scores aims to improve the therapeutic ratio of radiotherapy and to conceptualize personalized fractionation schemes. The new data regarding the involvement of the sixth "R" of radiobiology in HNSCC, especially for the HPV-driven subtype, but also for the "immune active" minority of HPV-negative HNSCCs, bring to the fore a multifactorial variation of the α/β ratio. The involvement of the antitumor immune response and the dose/fractionation/volume factors as well as the therapeutic sequence in the case of new multimodal treatments including immune checkpoint inhibitors (ICIs) could be included as an additional term in the quadratic linear formalism especially for hypo-fractionation regimens. This term should take into account the dual immunomodulatory effect (immunosuppressant and stimulator of antitumor immunity) of radiotherapy, which varies from case to case and can bring benefit or a detrimental effect.
... Defined as a set of genes (typically fewer than 100) whose expression covaries with a particular trait, certain gene expression signatures have already been incorporated into standard-of-care and clinical decision-making algorithms (e.g., OncotypeDx 5 , Mammaprint 6 ). In addition, signatures of radiosensitivity have been developed and have achieved level 1 evidentiary status for archival tissue [7][8][9][10] . ...
... Random forest models were built with the 'randomForest' package (version 4. [6][7][8][9][10][11][12][13][14], and each model grew 500 trees. All other parameters in training the prediction models were default. ...
Precision medicine offers remarkable potential for the treatment of cancer, but is largely focused on tumors that harbor actionable mutations. Gene expression signatures can expand the scope of precision medicine by predicting response to traditional (cytotoxic) chemotherapy agents without relying on changes in mutational status. We present a new signature extraction method, inspired by the principle of convergent phenotypes, which states that tumors with disparate genetic backgrounds may evolve similar phenotypes independently. This evolutionary-informed method can be utilized to produce consensus signatures predictive of response to over 200 chemotherapeutic drugs found in the Genomics of Drug Sensitivity in Cancer (GDSC) Database. Here, we demonstrate its use by extracting the Cisplatin Response Signature (CisSig). We show that this signature can predict cisplatin response within carcinoma-based cell lines from the GDSC database, and expression of the signatures aligns with clinical trends seen in independent datasets of tumor samples from The Cancer Genome Atlas (TCGA) and Total Cancer Care (TCC) database. Finally, we demonstrate preliminary validation of CisSig for use in muscle-invasive bladder cancer, predicting overall survival in a small cohort of patients who undergo cisplatin-containing chemotherapy. This methodology can be used to produce robust signatures that, with further clinical validation, may be used for the prediction of traditional chemotherapeutic response, dramatically increasing the reach of personalized medicine in cancer.
... Additionally, signatures of radiosensitivity have been developed and have achieved level 1 evidentiary status for archival tissue. [7][8][9][10] As seen in experimental and natural evolution, a variety of evolutionary trajectories can lead to the same phenotype. [11][12][13][14] Figure 1A shows a canonical example of convergent evolution, where genomically disparate species (bats and birds) both evolved the same phenotype of flight independently of one another. ...
... Random forest models were built with the 'randomForest' package (version 4. [6][7][8][9][10][11][12][13][14], and each model grew 500 trees. All other parameters in training the prediction models were default. ...
Precision medicine offers remarkable potential for the treatment of cancer, but is largely focused on tumors that harbor actionable mutations. Gene expression signatures can expand the scope of precision medicine by predicting response to traditional (cytotoxic) chemotherapy agents without relying on changes in mutational status. We present a novel signature extraction method, inspired by the principle of convergent evolution, which states that tumors with disparate genetic backgrounds may evolve similar phenotypes independently. This evolutionary-informed method can be utilized to produce signatures predictive of response to over 200 chemotherapeutic drugs found in the Genomics of Drug Sensitivity in Cancer Database. Here, we demonstrate its use by extracting the Cisplatin Response Signature, CisSig, for use in predicting a common trait (sensitivity to cisplatin) across disparate tumor subtypes (epithelial-origin tumors). CisSig is predictive of cisplatin response within the cell lines and clinical trends in independent datasets of tumor samples. Finally, we demonstrate preliminary validation of CisSig for use in muscle-invasive cancer, predicting overall survival in patients who undergo cisplatin-containing chemotherapy. This novel methodology can be used to produce robust signatures for the prediction of traditional chemotherapeutic response, dramatically increasing the reach of personalized medicine in cancer.
... Most of these factors are unknown before treatment and modified dynamically during the treatment course. Tumor radiosensitivity has been estimated before treatment using in vitro clonogenic assay (16)(17)(18) or a linear regression model derived from the specific gene expressions (19)(20)(21). However, these methods can only measure the tumor intrinsic cellular radiosensitivity and could not be utilized to assess intra-tumoral treatment dose-response modified by tumor cell repopulation (22)(23)(24), reoxygenation (25)(26)(27)(28), reactivation of immune response (29,30), etc. ...
Purpose:
Tumor voxel dose-response matrix (DRM) can be quantified using feedback from serial FDG-PET/CT imaging acquired during radiotherapy. This study investigated the dynamic characteristics and the predictive capability of DRM.
Methods:
FDG-PET/CT images were acquired before and weekly during standard chemoradiotherapy with the treatment dose 2 Gy × 35 from 31 head and neck cancer patients. For each patient, deformable image registration was performed between the pretreatment/baseline PET/CT image and each weekly PET/CT image. Tumor voxel DRM was derived using linear regression on the logarithm of the weekly standard uptake value (SUV) ratios for each tumor voxel, such as SUV measured at a dose level normalized to the baseline SUV0. The dynamic characteristics were evaluated by comparing the DRMi estimated using a single feedback image acquired at the ith treatment week (i = 1, 2, 3, or 4) to the DRM estimated using the last feedback image for each patient. The predictive capability of the DRM estimated using 1 or 2 feedback images was evaluated using the receiver operating characteristic test with respect to the treatment outcome of tumor local-regional control or failure.
Results:
The mean ± SD of tumor voxel SUV measured at the pretreatment and the 1st, 2nd, 3rd, 4th, and last treatment weeks was 6.76 ± 3.69, 5.72 ± 3.43, 3.85 ± 2.22, 3.27 ± 2.25, 2.5 ± 1.79, and 2.23 ± 1.27, respectively. The deviations between the DRMi estimated using the single feedback image obtained at the ith week and the last feedback image were 0.86 ± 4.87, -0.06 ± 0.3, -0.09 ± 0.17, and -0.09 ± 0.12 for DRM1, DRM2, DRM3, and DRM4, respectively. The predictive capability of DRM3 and DRM4 was significant (p < 0.001). The area under the curve (AUC) was increased with the increase in treatment dose level. The DRMs constructed using the single feedback image achieved an AUC of 0.86~1. The AUC was slightly improved to 0.94~1 for the DRMs estimated using 2 feedback images.
Conclusion:
Tumor voxel metabolic activity measured using FDG-PET/CT fluctuated noticeably during the first 2 treatment weeks and obtained a stabilized reduction rate thereafter. Tumor voxel DRM constructed using a single FDG-PET/CT feedback image after the 2nd treatment week (>20 Gy) has a good predictive capability. The predictive capability improved continuously using a later feedback image and marginally improved when two feedback images were applied.
... They concluded that their response predictor model based on gene expression could be very useful to improve the therapeutic approach of patients, assuming that the model would require in vivo validation [120]. Subsequently, Eschrich et al. extended the model to 48 cell lines from the NCI panel and included other biological variables, such as the mutational status of KRAS and TP53, as well as the tissue of origin [121,122]. Combining these data, they created a linear rank-based algorithm to calculate a radiosensitivity index (RSI). The RSI has since been validated in multiple cohorts of patients with different neoplastic entities (pancreas, glioblastoma, liver, brain and lung metastases, breast cancer) [123][124][125][126][127][128]. ...
Simple Summary
The identification of prognostic and predictive gene signatures of response to cancer treatment (radiotherapy) could help in making therapeutic decisions in patients affected by NSCLC. There are multiple proposals for gene signatures that attempt to predict survival or predict response to treatment (not radiotherapy), but they mainly focus on early stages or metastasis at diagnosis. In contrast, there have been few studies that raise these predictive and/or prognostic elements in nonmetastatic locally advanced stages, where treatment with ionizing radiation plays an important role. In this work, we review in depth previous works discovering the prognostic and predictive response factors in non-small cell lung cancer, specially focused on non-deeply studied radiation-based therapy.
Abstract
Non-small-cell lung cancer (NSCLC) is the leading cause of cancer-related death worldwide, generating huge economic and social impacts that have not slowed in recent years. Oncological treatment for this neoplasm usually includes surgery, chemotherapy, treatments on molecular targets and ionizing radiation. The prognosis in terms of overall survival (OS) and the different therapeutic responses between patients can be explained, to a large extent, by the existence of widely heterogeneous molecular profiles. The identification of prognostic and predictive gene signatures of response to cancer treatment, could help in making therapeutic decisions in patients affected by NSCLC. Given the published scientific evidence, we believe that the search for prognostic and/or predictive gene signatures of response to radiotherapy treatment can significantly help clinical decision-making. These signatures may condition the fractions, the total dose to be administered and/or the combination of systemic treatments in conjunction with radiation. The ultimate goal is to achieve better clinical results, minimizing the adverse effects associated with current cancer therapies.
... In contrast to the works discussed above, articles published on the identification of predictive or prognostic gene markers focused on cohorts whose main treatment was ionizing radiation are scarcer and practically nonexistent in the specific case of NSCLC. Torres-Roca and collaborators [67] identified genetic elements common to all neoplasms, which could explain the differences in radiosensitivity observed both in vitro and in clinical practice. The generation of the so-called "radiosensitivity index" (RSI) forms a predictive signature of response to radiotherapy treatment composed of the AR, cJUN, STAT1, PKC, RELA, ABCc, SUMO1, CDK1, HDAC1, and IRF1 genes, which has been subsequently validated in cohorts of patients with breast cancer, head and neck cancer, esophageal cancer, rectal cancer, and glioblastoma multiforme [50,67,68]. ...
... Torres-Roca and collaborators [67] identified genetic elements common to all neoplasms, which could explain the differences in radiosensitivity observed both in vitro and in clinical practice. The generation of the so-called "radiosensitivity index" (RSI) forms a predictive signature of response to radiotherapy treatment composed of the AR, cJUN, STAT1, PKC, RELA, ABCc, SUMO1, CDK1, HDAC1, and IRF1 genes, which has been subsequently validated in cohorts of patients with breast cancer, head and neck cancer, esophageal cancer, rectal cancer, and glioblastoma multiforme [50,67,68]. The work by Scott and collaborators [56] proposes a model to adapt the radiotherapy prescription to the individual sensitivity of the tumor of each patient. ...
Simple Summary
The search for prognostic and/or predictive gene signatures of the response to radiotherapy treatment can significantly aid clinical decision making. These signatures can condition the fractionation, the total dose to be administered, and/or the combination of systemic treatments and radiation. The ultimate goal is to achieve better clinical results, as well as to minimize the adverse effects associated with current cancer therapies. To this end, we analyzed the intrinsic radiosensitivity of 15 NSCLC lines and found the differences in gene expression levels between radiosensitive and radioresistant lines, resulting in a potentially applicable six-gene signature in NSCLC patients. The six-gene signature had the ability to predict overall survival and progression-free survival (PFS), which could translate into a prediction of the response to the cancer treatment received.
Abstract
Non-small-cell lung cancer (NSCLC) is the leading cause of cancer death worldwide, generating an enormous economic and social impact that has not stopped growing in recent years. Cancer treatment for this neoplasm usually includes surgery, chemotherapy, molecular targeted treatments, and ionizing radiation. The prognosis in terms of overall survival (OS) and the disparate therapeutic responses among patients can be explained, to a great extent, by the existence of widely heterogeneous molecular profiles. The main objective of this study was to identify prognostic and predictive gene signatures of response to cancer treatment involving radiotherapy, which could help in making therapeutic decisions in patients with NSCLC. To achieve this, we took as a reference the differential gene expression pattern among commercial cell lines, differentiated by their response profile to ionizing radiation (radiosensitive versus radioresistant lines), and extrapolated these results to a cohort of 107 patients with NSCLC who had received radiotherapy (among other therapies). We obtained a six-gene signature (APOBEC3B, GOLM1, FAM117A, KCNQ1OT1, PCDHB2, and USP43) with the ability to predict overall survival and progression-free survival (PFS), which could translate into a prediction of the response to the cancer treatment received. Patients who had an unfavorable prognostic signature had a median OS of 24.13 months versus 71.47 months for those with a favorable signature, and the median PFS was 12.65 months versus 47.11 months, respectively. We also carried out a univariate analysis of multiple clinical and pathological variables and a bivariate analysis by Cox regression without any factors that substantially modified the HR value of the proposed gene signature.
... Earlier work includes a 2012 study by Torres-Roca et al. where a radiosensitivity index (RSI) was developed which used the expression values of 10 genes and a linear regression model (20). This molecular signature was validated in BC cohorts where it strati ed patients as radioresistant or radiosensitive, and where radiosensitive patients had an improved 5-year relapse-free survival compared to radioresistant patients (95% vs. 75%) that was not observed in RT-untreated patients (21). ...
Background
Radiation therapy (RT) is frequently recommended for post-surgery treatment of early-stage breast cancer (BC) patients, though not all benefit. Clinical factors currently guide RT treatment decisions. At present, models to predict RT-benefit predominantly use statistical methods with modest performance. In this paper we present a high-accuracy genomic Machine Learning (ML) model to predict RT-benefit in early-stage BC patients. We also present a novel method for selecting genomic features for training ML algorithms.
Methods
Gene expression data from 463 early-stage BC patients treated with surgery and RT from the METABRIC cohort were obtained. Wilcoxon Rank Sum (Wilcoxon RS) test and Cox Proportional Hazards (Cox PH) were used to reduce the number of genes used to train eight ML algorithms. ML algorithms were trained on 80% of data using 10-fold cross validation and tested on 20% of data to assess performance in predicting relapse status.
Results
Genome-wide gene expression data was reduced by 96% using Wilcoxon RS and Cox PH to a 1,596 gene set and a 977 gene set. These gene sets were used to train eight ML algorithms resulting in models that ranged in performance accuracies from 54.01% to 95.6%. Highest accuracies were obtained using Support Vector Machine (SVM977–93.41%, SVM1596–95.6%) and Neural Networks algorithms (NN977 – 92.31%, NN1596 – 93.41%). In RT-untreated patients, accuracies of all models were 30% to 40% lower compared to RT-treated patients. SVM977 had the highest sensitivity of 91.09%. Members of the 977 set were enriched with genes involved in cell cycle and differentiation as well as genes associated with radiosensitivity and radioresistance.
Conclusion
This study presents a novel genomic feature selection approach that used Wilcoxon RS followed by Cox PH to reduce the number of genes from genome-wide gene expression data used for training ML algorithms by 96%. This approach led to an SVM model that used the expression values of 977 genes to predict RT-benefit in early-stage BC patients with 93.41% accuracy. This work demonstrates that ML models can be clinically useful for predicting cancer patient outcomes.
