Optimizing carboplatin dosing using Computed Tomography derived body composition and serum creatinine

Carboplatin is an anticancer drug used for the treatment of various types of
cancer including non-small cell lung cancer (NSCLC), small cell lung cancer
(SCLC), ovarian cancer, and breast cancer. It is mostly secreted by the kidney
with its clearance being linearly correlated with the glomerular filtration
rate (GFR). Therefore, the dosing of carboplatin is based on the GFR, and the
target exposure is expressed as the target Area Under the Curve (AUC) using the
Calvert formula. In clinical practice, the GFR is substituted by the estimated
creatinine clearance (CrCl) using the Cockcroft-Gault (CG) formula. CrCL is
determined by the difference between creatinine production (i.e. muscle mass)
and serum creatinine concentration. The CG formula uses surrogate markers of
muscle mass (i.e. age, gender, weight) and serum creatinine to estimate CrCl.
However, the CG formula is not well suited for extreme body compositions, often
seen in cancer patients, and leads to overprediction in overweight and obese
patients or cachectic patients with low serum creatinine values. This can lead
to increased risk of (serious) side effects and hospitalization, decreased
quality of life, postponement of treatment, and dose reduction. On the other
hand, can the CG formula lead to underprediction in underweight patients
resulting in potentially inadequate treatment.

A better, specific, and more direct way to measure a patient's muscle mass can
be done using a CT-scan already performed in standard-of-care. Recent advances
in deep learning and medical imaging have made it possible to utilize regular
CT-scans of the skeletal muscle cross-sectional area of lumbar 3 to predict the
total muscle mass of a patient accurately. This makes it possible to measure
creatinine clearance directly using CT-scans and to predict carboplatin
clearance. Subsequently, by estimating carboplatin clearance, a more precise
carboplatin dosing can be established, preventing potentially under- or
overdosing of carboplatin, especially in patients with extreme body
composition.

The overall objective of this study is to develop and evaluate an optimized
dosing algorithm (CT-CL) for carboplatin using serum creatinine and CT-derived
body composition that can be easily implemented in clinical practice.

This study consists of two parts:
1. PART A is a prospective observational pharmacokinetic study to asses
carboplatin pharmacokinetics in relation to CT-derived body composition and
serum creatinine and subsequently develop an optimized dosing algorithm.
2. PART B is a simulation study to evaluate the developed dosing algorithm.


Patients are treated for their cancer according to routine clinical practice,
standard protocols, and treatment regimens. The dose of carboplatin to be
administered will be calculated using the Calvert formula. In addition, a
CT-scan including L3 serum creatinine at baseline is used to assess the body
composition - both are collected as standard-of-care. Both the abdominal and
the thoracal CT-scan include L3. For each patient, five blood samples will be
taken on day 1 of the first cycle of carboplatin treatment.

Next, a virtual cohort of 1000 patients will be simulated with NONMEM V7.4
using Monte Carlo simulations. This population will be generated using the
virtual human population generator PopGen (36). For the simulation of the
covariates, the median plus distribution found in the study population from the
PK analysis will be used. Patients will subsequently be dosed according to the
Calvert formula and CT-CL. The simulated AUC will be evaluated on the bias
(MPE%), precision (MAPE%), and accuracy (RMSE) relative to the target AUC.
Lastly, the percentage of patients within 90-110% of the target AUC (± 10%)
will be assessed.

All patients will be treated according to standard protocols and treatment
regimens. We consider the extra burden of participating in this study limited.
The extra interventions, compared to routine care, consist of blood sampling.
To ensure the minimal impact of study participation on daily life, we will use
a limited sampling strategy of only five blood samples. For each patient five
PK samples will be taken on day 1 of cycle 1 of carboplatin treatment for a
total of 25 mL. This is a minimal amount compared to the total blood volume of
a patient. There is a minimal risk that the venflon may cause slight irritation
or thrombophlebitis. All other visits, CT-scan at baseline, and laboratory
investigations are performed in standard-of-care.

The aim of this study is to develop and evaluate an optimized dosing algorithm
for carboplatin using serum creatinine and CT-derived body composition. The
results of this study can improve future treatment of patients with
carboplatin, especially with abnormal body composition.

The results of our study can improve future treatment of patients with
carboplatin, especially with abnormal body composition, and will help switch
from one-size-fits-all formulas to more personalized medicine.