Our study demonstrated that patients with solid tumor malignancies and hypoalbuminemia had significantly shorter treatment time on highly protein-bound oral TKIs. To our knowledge, this is the first study to show that patients’ baseline albumin levels as a continuous variable were directly correlated with their treatment duration solely due to tolerability of these drugs, while accounting for important covariates. There are two recently published single-center retrospective studies with similar objectives of assessing oral oncolytics tolerability in patients with solid tumors and hypoalbuminemia. One by Marini and colleagues evaluating the tolerability of 16 oral TKIs and the other by Murdock and colleagues evaluating 14 oral TKIs and two antiandrogens, namely abiraterone and enzalutamide.14,21
In the study by Murdock and colleagues, 309 patients were screened and 143 were included, 42% of whom had hypoalbuminemia at baseline and 75% were male. In their primary outcome, authors report a median time to medication discontinuation from documented toxicities of 22 months in the hypoalbuminemia group vs. not reached in patients without hypoalbuminemia. The covariates in their Cox Regression analyses included history of liver dysfunction, line of therapy, baseline ECOG performance status, and presence of hypoalbuminemia, however neither renal function nor severity of toxicities were reported. In the study conducted by Marini and colleagues, a total of 220 patients were included, however excluded number of patients and concurrent intravenous oncolytics use are unknown. 45.9% of included patients had hypoalbuminemia and 55% were male. Although their primary outcome was all-cause medication discontinuation, they also reported rates and time to treatment discontinuation due to intolerability as secondary outcomes which was 37% and 96 days in the hypoalbuminemia group vs 16% and 288 days in those without hypoalbuminemia respectively (P < 0.001 for both outcomes). In addition to liver disease, Marini and colleagues also reported on presence or absence of proteinuria in patients but only adjusted for performance status, previous lines of therapy, metastases, and liver disease covariates in their Cox Regression analyses.
Notably, our study has some similarities and differences to the aforementioned studies. We screened 1,011 patients, 282 of whom met inclusion criteria for discontinuation solely due to medication intolerability (all exclusion criteria are reported in Fig. 1). The reason for our rigorous selection process was to minimize any confounding variables that were not a true reflection of TKI discontinuation due to intolerability. In our study, 48% of patients had hypoalbuminemia and 51.8% were male, which further corroborates previous study findings. In addition, our median time to treatment discontinuation due to AEs were 2.8 months in low vs 4.3 in normal albumin groups, similar to Marini and colleagues’ findings. Despite these similarities, our study has some major differences. First, we included variables that can impact free albumin levels such as total bilirubin and corrected calcium levels, nutrition marker such as BMI, and renal function such as serum creatinine that can widely affect the pharmacokinetics and clearance of these drugs, all as continuous data in our Cox Regression analyses. Whereas neither study reported these values nor included them in their analyses and reported liver dysfunction or proteinuria as dichotomized data. Furthermore, contrary to two previous studies, our study is the first to report hospitalization and all-cause mortality rates as well as severity of AEs. From our univariate analyses, we found that patients in the hypoalbuminemia group were found to have higher corrected calcium levels and higher all-cause mortality rates (Table 2). These results could perhaps be explained by having higher number of patients with stage IV disease in the hypoalbuminemia group as cancer-related hypercalcemia is often associated with disseminated disease and poor prognosis. In addition, higher BMI values were associated with longer time on TKI treatment, whereas higher serum creatinine levels were associated with shorter time on treatment. More patients in the normal albumin group experienced grade 1/2 AEs vs more patients in the hypoalbuminemia group experienced grade 3/4 AEs, all of which are new findings compared to two previous studies. Our finding are, however, in confirmation with previous research evaluating the association between malnutritional factors, such as BMI or albumin levels, with patients’ chemotherapy tolerance or survival outcomes.4,22 Additionally, several studies have shown the nephrotoxic AEs and baseline renal impairment effects on the pharmacokinetics and toxicities of oral TKIs, which corroborates our findings in this study.23,24 Conversely, patients’ total bilirubin levels had no impact on their treatment duration, which is in parallel with Murdock and colleagues’ findings regarding their multivariate analysis on the effects of liver dysfunction or cirrhosis history and time to drug discontinuation from toxicity.
When looking at impact of dose reduction on time on treatment, we found that dose reduction increased total time on treatment in the normal albumin group only. This new finding may signify that certain patients can potentially benefit from TKI dose reduction by increasing their total time on treatment without experiencing severe toxicities or disease progression. However, our multivariate analysis showed patients who started their TKIs at a reduced dose did not have a longer time on treatment compared to those who started at full doses. However, only 27 patients total started at a reduced dose and a bigger sample size may be needed to assess the effects of empiric dose reduction on treatment duration.
Not only did our study show that patients with lower albumin levels had shorter time on treatment due to experiencing toxicities, but our findings also showed that patients with hypoalbuminemia experienced more severe adverse events (grade 3/4) compared to those in the normal albumin group. The reported AEs in this study were in line with known AEs of each TKI drug class previously reported in studies.16 The most common AEs are reported in Fig. 4, with diarrhea occurring most frequently with EGFR inhibitors, fatigue with ALK and VEGF inhibitors, and rash and itching with MEK inhibitors.
There are a variety of factors that make cancer care complex such as the pathophysiology of cancer itself, patients’ comorbidities, and the high-risk nature of cancer medications. Although, we aimed to take many of these confounding variables into consideration when designing our study, there may still be other contributing factors that can impact patients’ tolerability of oral TKIs. This is a retrospective study, where all data collection relies on manual chart review, which may pose bias into the study or thorough and accurate documentation may be lacking. The single-center nature of this study also limits the external validity of results. For instance, 95.5% of patients in this study were Caucasians and representation of other races is lacking. Additionally, many cancer patients take multiple concomitant medications that may impact TKIs’ drug concentrations whether through drug-drug interactions or protein binding displacement that we did not account for with the exception of excluding intravenous oncolytics.