Prospective Validation of Machine Learning Model to Predict Platinum Induced Nephrotoxicity in Cancer Patients

Recruiting

• Conditions: Chemotherapy Side Effects; Machine Learning; Acute Kidney Injury; Acute Kidney Disease

• Registration Number: NCT07114276
• Lead Sponsor: Taipei Medical University

Brief Summary

This study aims to investigate the utility of predictive models for chemotherapy-induced nephrotoxicity in the Taiwanese cancer population.

We will prospectively collect clinical data from enrolled patients, including demographic information, comorbidities, laboratory data, and chemotherapy treatment details. After chemotherapy administration, we will monitor patients' renal function over time to assess the development of nephrotoxicity, based on changes in serum creatinine and other relevant clinical criteria.

The primary objective is to evaluate and compare the predictive performance of the machine learning model and clinical physicians. This will be done using the area under the receiver operating characteristic curve (AUROC) as the main metric for discrimination performance.

Detailed Description

This is a prospective cohort study conducted at Wan Fang Hospital, designed to evaluate the predictive performance and clinical utility of pre-developed machine learning models for detecting nephrotoxicity in cancer patients receiving platinum-based chemotherapy. The models were developed using a Long Short-Term Memory (LSTM) architecture, trained on a retrospective dataset from January 1, 2009 to January 31, 2022. All model parameters were locked after training to ensure reproducibility, prevent data leakage, and maintain the integrity of prospective validation.

We recruit cancer patients who receive platinum-containing agents (such as Cisplatin and Carboplatin) from October 2023 to August 2025 and follow-up to November 2025. After confirming eligibility based on inclusion and exclusion criteria, we obtain written informed consent from each participant prior to data collection.

Each administration of platinum chemotherapy is treated as a separate prediction case. For every administration, we aim to predict two renal outcomes:

1. Acute Kidney Injury (AKI) occurring within 14 days after the chemotherapy. Based on CTCAE 5.0

2. Acute Kidney Disease (AKD) occurring within 89 days after the chemotherapy. Based on ADQI (Acute Disease Quality Initiative 2016)

After obtaining the patients' informed consent, we will collect the patient's general information (gender, age, height, weight), cancer-related information (stage), administration situation (administration date, administration course, dosage, concomitant chemotherapy drugs, the number of administrations), laboratory data (serum creatinine, glomerular filtration rate). The relevant follow-up data will be used as the therapeutic effect of disease drugs and disease prognosis. Data will be presented under the research number.

In addition to model prediction, we will obtain clinicians' risk assessments for the same chemotherapy administrations. There are four physicians participating in the study. Each chemotherapy case (i.e., each drug administration) will be randomly assigned to one of the four physicians for independent prediction. The physician will be blinded to model predictions and required to return their judgment regarding the risk of AKI and AKD within one week.

If the interval between two chemotherapy administrations exceeds 42 days (6 weeks), or if there is a change in the chemotherapy regimen-for example, switching from Cisplatin to Carboplatin, or vice versa-the patient will be treated as a new subject with a separate research ID for subsequent analysis.

We predict the occurrence of acute kidney injury and acute kidney disease in each patient with clinical features with the existing model and compare model predictions with clinicians' predictions. The performance of models and clinicians is evaluated with machine learning metrics like AUROC, AUPRC, sensitivity, etc. We tend to classify the patients into several nephrotoxicity risk groups based on models' and clinicians' predictions. We expect to verify the utility of the model and analyze the causes of the disparities between predicted outcomes and correct values.

Categorical variables were analyzed using the Chi-square test or Fisher's exact test based on the expected cell counts. Continuous variables were compared using the independent t-test or the Mann-Whitney test, depending on the distribution of the data. A two-sided p-value \< 0.05 was considered statistically significant. The 95% confidence intervals (CIs) for AUROC and AUPRC were calculated using DeLong's test and the bootstrap method with 1000 iterations. All analyses were conducted using Python (version 3.11.12) and R (version 4.4.3).

Study Timeline

• Study Start: 2023-10-30
• Status Verified: 2025-08
• Study First Submitted: 2025-08-04
• Study First Submitted QC: 2025-08-04
• Last Update Submitted: 2025-08-04
• Study First Posted: 2025-08-11
• Last Update Posted: 2025-08-11
• Primary Completion: 2025-11-30
• Study Completion: 2026-01-31

Recruitment & Eligibility

• Status: RECRUITING
• Sex: All
• Target Recruitment: 110

Inclusion Criteria

• Patients under clinical diagnosis of cancer with treatments including at least taking one course treatment of Cisplatin and Carboplatin from Dec,2022 to July,2026, at least having one serum creatinine data before and after the administration, willing to provide DNA sample and sign the informed consent will be recruited.

Exclusion Criteria

• Patients who are young than 20 years old or older than 89 years old, pregnant women, infected by Human Immunodeficiency Virus (HIV), administered by Ifosfamide, couldn't evaluate their kidney function, refuse to provide DNA sample and sign the informed consent will be excluded.

Study & Design

• Study Type: OBSERVATIONAL
• Study Design: Not specified

Primary Outcome Measures

Name	Time	Method
AUROC comparison	89 days	Comparison of the area under the receiver-operator characteristic (ROC) curves between the predictions made by the machine learning models and by clinicians, to predict AKI within 14 days and AKD within 89 days

Secondary Outcome Measures

Name	Time	Method
Incidence and odd ratios in each risk level group	89 days	Ensure the incidence of nephrotoxicity in each risk level group

Trial Locations

Locations (1)

Taipei Medical University Wan Fang Hospital; 🇨🇳 Taipei, Taiwan