Lyndsey F. Meyer, Cynthia J. Musante, Theodore R. Rieger
Pfizer Inc.
Background: Unintentional weight loss is a symptom of serious disease or illness. Two widely recognized, yet, under diagnosed diseases that cause significant unintentional weight loss include cancer-associated cachexia and malnutrition of aging. Both diseases lead to muscle weakness, fatigue, poor quality of life, and higher mortality rates. Up to 30% of cancer patients experience cancer-associated cachexia [1] and 25% of people over 65 years old have unintentional weight loss with no obvious underlying cause [2]. Despite its prevalence, there are few drugs to address unintentional weight loss in these populations, such as megestrol acetate, an appetite stimulant and prostaglandin mimetic. Among the challenges for developing therapies against cachexia/unintentional weight loss is a limited number of clinical trials have evaluated the time-course of weight loss in patients with cancer cachexia or malnutrition associated with aging.
Objectives: The objective of this work was to aid in clinical trial design by predicting weight gain responses in a virtual patient population to a potential therapy for either cancer-associated cachexia or malnutrition of aging. To generate virtual patients, we employed available body weight data from real-world databases, such as electronic health records, to parameterize a QSP body composition model [3].
Methods:
De-identified patient data was obtained from the Optum Market Claims Database for patients with non-small cell lung cancer (NSCLC) or Point Click Care databases to obtain patient data from skilled nursing facilities across North America. To identify NSCLC patients that would be most likely to enroll in a clinical trial, we extracted patients receiving megestrol acetate for weight loss prevention and propensity score matched [4] a cohort of patients to represent a naïve patient population (N = 449). Propensity score matching was done based on age, sex, BMI, percentage of weight loss from cancer to cachexia diagnosis. For patients with malnutrition of aging, serial measures of body weight were extracted by identifying patients over 65 years old with weight loss greater than 2 kg, no history of cancer diagnosis, and the ability to provide consent based on information from the minimum safety dataset questionnaire.
To understand the time-course of disease progression for each clinical trial population, we generated plausible virtual patients based on the real-world data using a QSP model of body composition with a Metropolis-Hastings algorithm. Virtual patients were included for clinical trial simulation if they had weight loss trajectories that were a close match to the observed real-world data. Following VP selection, we simulated the effects of potential therapies by reversing the caloric deficit to optimize clinical trial duration for a target change in body weight.
Results: Using modeling and simulation, we estimated the variability in both the rate and extent of weight loss in patients with unintentional weight loss. The weight loss trajectory agreed with available published clinical trial data [5-7] such that the clinical data falls within the 90% prediction interval. The model estimated patients have a caloric deficit of between 20% to 40% below their baseline caloric needs. Finally, we demonstrated with simulations that reversing the effects of a caloric deficit can improve the trajectory of weight loss, and in some patients, result in weight gain which is linked with improved survival and quality of life [8].
Conclusions: The data informed model predicts attenuation of weight loss if the caloric deficit is overcome, which is on average ~230 kcal/day. Leveraging real-world data allowed us to fill gaps in knowledge about the time-course of unintentional weight loss progression and to prospectively simulate therapies to enable optimal clinical trial design.
References:
- S. von Haehling, et al. J Cachexia Sarcopenia Muscle, 2010. 1(1): p. 1-5
- Shabbir M.H. Alibhai, Greenwood, and Payette. CMAJ, 2005.172(6): p. 773–780.
- Hall, K.D. Am J Physiol Endocrinol Metab, 2010. 298(3): p. E449-66.
- P. C. Austin. Multivariate Behav Res 2011. 46(3): p. 399-424
- Temel, J.S., et al. Lancet Oncol, 2016. 17(4): p. 519-31.
- Takayama, K., et al. Support Care Cancer, 2016. 24(8): p. 3495-505.
- Katakami, N., et al. Cancer, 2018. 124(3): p. 606-16.
- Davidson, W., et al. Clin Nutr, 2004. 23(2): p. 239-47.
Reference: PAGE 32 (2024) Abstr 10889 [www.page-meeting.org/?abstract=10889]
Poster: Real-world data (RWD) in pharmacometrics