Dataset Information

Random Forest for Predicting Treatment Response to Radioiodine and Thyrotropin Suppression Therapy in Patients With Differentiated Thyroid Cancer But Without Structural Disease.

ABSTRACT:

Background

We aimed to develop a machine-learning model for predicting treatment response to radioiodine (131I) therapy and thyrotropin (TSH) suppression therapy in patients with differentiated thyroid cancer (DTC) but without structural disease, based on pre-treatment information.

Patients and methods

Overall, 597 and 326 patients with DTC but without structural disease were randomly assigned to "training" cohorts for predicting treatment response to 131I therapy and TSH suppression therapy, respectively. Six supervised algorithms, including Logistic Regression, Support Vector Machine, Random Forest (RF), Neural Networks, Adaptive Boosting, and Gradient Boost, were used to predict effective response (ER) to 131I therapy and biochemical remission (BR) to TSH suppression therapy.

Results

Stimulated and suppressed thyroglobulin (Tg) and radioiodine uptake before the current course of 131I therapy were mostly attributed to ER to 131I therapy, while thyroid remnant available on the post-therapeutic whole-body scan at the last course of 131I therapy and TSH were greatly contributed to Tg decline under TSH suppression therapy. RF showed the best performance among all models. The accuracy and area under the receiver operating characteristic curve (AUC) for segregating ER from non-ER during 131I therapy with RF were 81.3% and 0.896, respectively. The accuracy and AUC for predicting BR to TSH suppression therapy with RF were 78.7% and 0.857, respectively.

Conclusion

This study demonstrates that machine learning models, especially the RF algorithm are useful tools that may predict treatment response to 131I therapy and TSH suppression therapy in DTC patients without structural disease based on pre-treatment routine clinical variables and biochemical markers.

SUBMITTER: Sa R

PROVIDER: S-EPMC10769791 | biostudies-literature | 2024 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Random Forest for Predicting Treatment Response to Radioiodine and Thyrotropin Suppression Therapy in Patients With Differentiated Thyroid Cancer But Without Structural Disease.

Sa Ri R Yang Taiyu T Zhang Zexu Z Guan Feng F

The oncologist 20240101 1

<h4>Background</h4>We aimed to develop a machine-learning model for predicting treatment response to radioiodine (131I) therapy and thyrotropin (TSH) suppression therapy in patients with differentiated thyroid cancer (DTC) but without structural disease, based on pre-treatment information.<h4>Patients and methods</h4>Overall, 597 and 326 patients with DTC but without structural disease were randomly assigned to "training" cohorts for predicting treatment response to 131I therapy and TSH suppress ...[more]

PMID: 37669005

Similar Datasets

Project description:BackgroundThe diagnosis of radioiodine refractory differentiated thyroid cancer (RAIR-DTC) is primarily based on clinical evolution and iodine uptake over the lesions, which is still time-consuming, thus urging a predictive model for timely RAIR-DTC informing. The aim of this study was to develop a nomogram model for RAIR prediction among DTC patients with distant metastases (DM).MethodsData were extracted from the treatment and follow-up databases of Peking Union Medical College Hospital between 2010 and 2021. A total of 124 patients were included and divided into RAIR (n=71) and non-RAIR (n=53) according to 2015 ATA guidelines. All patients underwent total thyroidectomy followed by at least two courses of RAI treatment. Serological markers and various clinical, pathological, genetic status, and imaging factors were integrated into this study. The pre-treatment stimulated Tg and pre- and post-treatment suppressed Tg at the first and second course RAI treatment were defined as s-Tg1, s-Tg2, sup-Tg1, and sup-Tg2, respectively. Δs-Tg denoted s-Tg1/s-Tg2, and Δs-TSH denoted s-TSH1/s-TSH2. Multivariate logistic regression and correlation analysis were utilized to determine the independent predictors of RAIR. The performance of the nomogram was assessed by internal validation and receiver operating characteristic (ROC) curve, and benefit in clinical decision-making was assessed using decision curve.ResultsIn univariate logistic regression, nine possible risk factors were related to RAIR. Correlation analysis showed four of the above factors associated with RAIR. Through multivariate logistic regression, Δs-Tg/Δs-TSH<1.50 and age upon diagnosis were obtained to develop a convenient nomogram model for predicting RAIR. The model was internally validated and had good predictive efficacy with an AUC of 0.830, specificity of 0.830, and sensitivity of 0.755. The decision curve also showed that if the model is used for clinical decision-making when the probability threshold is between 0.23 and 0.97, the net benefit of patients is markedly higher than that of the TreatAll and TreatNone control groups.By using 1.50 as a cut-off ofΔs-Tg/Δs-TSH, differing biochemical progression among the generally so-called RAIR can be further stratified as meaningfully rapidly or slowly progressive patients (P=0.012).ConclusionsA convenient user-friendly nomogram model was developed with good predictive efficacy for RAIR. The progression of RAIR can be further stratified as rapidly or slowly progressive by using 1.50 as a cut-off value of Δs-Tg/Δs-TSH.

Project description:Objective This study aimed to establish and validate the nomograms for predicting overall survival (OS) probabilities in differentiated thyroid cancer (DTC) patients who received and did not receive radioiodine therapy (RAI), respectively. Methods In this study, 11, 099 patients diagnosed with DTC in the Surveillance, Epidemiology, and End Results (SEER) database from 2004 to 2016 were selected. Whether they have RAI, they are divided into RAI (n=6427) and non-RAI (n=4672) groups. They were randomly assigned to either a training cohort (RAI: n=4498, non-RAI: n=3263) or a validation cohort (RAI: n=1929, non-RAI: n=1399) using R software to divide the patients in a 7-to-3 ratio randomly. Variables were selected using a backward stepwise method in a Cox regression model to determine the independent prognostic factors, which were then utilized to build two nomograms to predict the 5-, 8-, and 10-year OS probabilities in DTC patients with or without RAI. The concordance index (C‐index), the area under the time-dependent receiver operating characteristics curve (AUC), the net reclassification improvement (NRI), the integrated discrimination improvement (IDI), calibration plotting, and decision-curve analysis (DCA) were used to evaluate the performance of our models. Results The multivariate analyses demonstrated that birth of the year, race, histological type, tumor size, grade, TNM stage, lymph node dissections, surgery, and chemotherapy were risk factors for OS. Compared to the AJCC stage, the C‐index (RAI: training group: 0.911 vs. 0.810, validation group: 0.873 vs. 0.761; non-RAI: training group: 0.903 vs. 0.846, validation group: 0.892 vs. 0.808). The AUC values for the training cohort (RAI: 0.940, 0.933, and 0.942; non-RAI: 0.891, 0.884, and 0.852 for the 5-, 8-, and 10-year OS, respectively) and validation cohort (RAI: 0.855, 0.825, and 0.900, non-RAI: 0.867, 0.896, and 0.899), and the calibration plots of both two models all exhibited better performance. Additionally, the NRI and IDI further showed that they exhibited good 5-, 8-, and 10-year net benefits. Conclusion We have established the prediction models of DTC patients with or without RAI respectively through various variables. The nomogram may be more targeted to guide clinical decisions in the future.

Project description:BackgroundLevothyroxine suppression of thyrotropin (TSH) is broadly applied to patients with thyroid cancer despite lack of consensus on the optimal TSH concentration necessary to reduce cancer recurrence while minimizing toxicity from subclinical hyperthyroidism. The objectives of this study were to examine the beneficial effects and the cardiac and skeletal toxicity of TSH suppression in well-differentiated thyroid carcinoma (DTC).MethodsA total of 771 patients (569 women) at ATA low or intermediate risk of recurrence, with a mean age of 48±14 years, and undergoing total thyroidectomy at a tertiary care center between 2000 and 2006 were followed for a median of six and a half years. They were divided into a suppressed TSH group (median TSH ≤0.4 mIU/L) and a nonsuppressed group (median TSH >0.4 mIU/L). Structural recurrence of thyroid cancer, postoperative atrial fibrillation (AF), and osteoporosis were examined in the two groups. Osteoporosis was only examined in women.ResultsA total of 43/771 (5.6%) patients recurred, 29/739 (3.9%) patients were diagnosed with postoperative osteoporosis, and 17/756 (2.3 %) were diagnosed with postoperative AF. Despite similar rates of recurrence (HR 1.02, p=0.956 [CI 0.54-1.91]), patients treated to a median TSH ≤0.4 mIU/L were at increased postoperative risk of a composite outcome of AF and osteoporosis (HR 2.1, p=0.05 [CI 1.001-4.3]) compared to those not suppressed. A differential risk of AF alone (HR 0.78, p=0.63 [CI 0.3-2.1]) was not detected, but postoperative osteoporosis was increased among women with a suppressed TSH compared to those not suppressed (HR 3.5, p=0.023 [CI 1.2-10.2]). The increased risk of postoperative osteoporosis disappeared when the patient's median TSH was maintained around 1 mIU/L.ConclusionTSH suppression significantly increases the risk of postoperative osteoporosis without changing tumor recurrence in ATA low- and intermediate-risk patients with DTC. Future interventions should focus on avoiding harm in indolent disease.

Project description:Importance:Suppression of thyrotropin (often referred to as thyroid-stimulating hormone, or TSH) with levothyroxine used in management of intermediate- and high-risk differentiated thyroid cancer (DTC) to reduce the likelihood of progression and death is based on conflicting evidence. Objective:To examine a cohort of patients with intermediate- and high-risk DTC to assess the association of thyrotropin suppression with progression-free survival (PFS) and overall survival. Design, Setting, and Participants:This cohort study used a multicenter database analysis including patients from tertiary referral centers and local clinics followed up for a mean (SD) of 7.2 (5.8) years. Patients with DTC treated uniformly with total thyroidectomy and radioactive iodine between January 1, 1979, and March 1, 2015, were included. Among the 1012 patients, 145 patients were excluded due to the lack of longitudinal thyrotropin measurements. Exposures:Levothyroxine therapy to target thyrotropin suppression with dose adjustments based on changing thyrotropin goal. Main Outcomes and Measures:The primary outcome measures were overall survival and PFS. A Cox proportional hazards model was used to assess the contribution of age, sex, tumor size, histology, and lymph node and distant metastases at landmarks 1.5, 3.0, and 5.0 years. The patients were divided into 3 groups based on mean thyrotropin score before each landmark: (1) suppressed thyrotropin, (2) moderately suppressed or low-normal thyrotropin, and (3) low-normal or elevated thyrotropin. Results:Among 867 patients (557 [64.2%] female; mean [SD] age, 48.5 [16.5] years) treated with a median (range) cumulative dose of 151 (30-1600) mCi radioactive iodine, disease progression was observed in 293 patients (33.8%), and 34 patients (3.9%) died; thus, the study was underpowered in death events. Thyrotropin suppression was not associated with improved PFS at landmarks 1.5 (P = .41), 3.0 (P = .51), and 5.0 (P = .64) years. At 1.5 and 3.0 years, older age (hazard ratio [HR], 1.06; 95% CI, 1.03-1.08 and HR, 1.05; 95% CI, 1.01-1.08, respectively), lateral neck lymph node metastases (HR, 4.64; 95% CI, 2.00-10.70 and HR, 4.02; 95% CI, 1.56-10.40, respectively), and distant metastases (HR, 7.54; 95% CI, 3.46-16.50 and HR, 7.10; 95% CI, 2.77-18.20, respectively) were independently associated with subsequent time to progression, while at 5.0 years, PFS was shorter for patients with lateral neck lymph node metastases (HR, 3.70; 95% CI, 1.16-11.90) and poorly differentiated histology (HR, 71.80; 95% CI, 9.80-526.00). Conclusions and Relevance:Patients with intermediate- and high-risk DTC might not benefit from thyrotropin suppression. This study provides the justification for a randomized trial.

Project description:BackgroundOutcomes vary among patients with radioiodine refractory (RR) differentiated thyroid cancer (DTC). The prognostic factors for survival are not well-known, resulting in difficulty in selecting patients for new targeted therapies. We assessed overall survival (OS) and cancer-specific survival (CSS) from RR-DTC to identify prognostic factors associated with survival.Patients and methodsThe data on all cases of metastatic RR-DTC treated in our center from 1990 to 2011 were retrospectively reviewed. Survival was estimated using the Kaplan-Meier method; associated prognostic factors were assessed using Cox's model.ResultsOf 153 cases of metastatic DTC, 59% (n = 91) met a criterion for RR: that is, 60% (n = 55) had at least 1 metastasis without (131)I uptake; 21% (n = 19) had progressive disease (PD) despite (131)I; 19% (n = 17) had persistent disease despite a cumulative activity of (131)I of ≥600 mCi. After the diagnosis of RR, median OS was 8.9 years (95% confidence interval [CI]: 5.4-NR); median CSS was 9.6 years (95% CI: 6.01-NR). In multivariate analyses, PD despite (131)I as a criterion for RR disease and the time from initial diagnosis of DTC to diagnosis of RR <3 years were the only independent prognostic factors for poor OS and CSS. Thyroglobulin doubling time (Tg-DT) was assessed in 31 of 91 cases. Among the 11 patients with Tg-DT for <1 year or undetectable Tg, 6 deaths occurred, whereas only 3 died of 20 patients with Tg-DT >1 year or negative Tg-DT.ConclusionThe identification of prognostic factors for decreased survival in RR-DTC may improve the selection of patients for targeted agents.Implications for practiceThis study shows a great heterogeneity in terms of prognosis in radioiodine refractory differentiated thyroid carcinoma. Poorer prognosis is observed in patients with tumor progression or with a diagnosis of radioiodine resistance within 3 years after the initial diagnosis of thyroid cancer. Those findings could lead to improvements in the selection of patients for targeted therapies.

Dataset Information

Random Forest for Predicting Treatment Response to Radioiodine and Thyrotropin Suppression Therapy in Patients With Differentiated Thyroid Cancer But Without Structural Disease.

Background

Patients and methods

Results

Conclusion

Publications

Random Forest for Predicting Treatment Response to Radioiodine and Thyrotropin Suppression Therapy in Patients With Differentiated Thyroid Cancer But Without Structural Disease.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets