The present study enrolled a large sample of 41,911 patients with diverse cancer and demonstrated that higher serum creatinine fluctuation during hospital admission was associated with a higher mortality risk within one year. Notably, fluctuation of creatinine was associated with mortality regardless of whether all the values were within the clinically normal range or not. Furthermore, stratified analyses according to sex, age, measurement times of creatinine, and whether had undergone chemoradiotherapy or surgery did not significantly change the results. Sensitivity analyses also confirm the robustness of the conclusion.
In previous studies, researchers evaluated the predictive value of absolute or relative changes in creatinine on mortality and illustrated that changes in serum creatinine identified the patients who were at increased risk for short- and long-term death.29–36 Our findings extend and clarify those of previous studies and extrapolated the conclusion to patients with cancer. Moreover, some existing observations reached conflicting conclusions. For example, they found that cancer patients are at higher risk of death with increases in serum creatinine or AKI.35,37–40 In contrast, an increase in serum creatinine was not associated with hematological adverse events in patients with breast cancer.36 In addition, one study based on the UK biobank illustrated that creatinine-cystatin C was associated with neither cancer incidence nor the risk of cancer death. 3
Previous findings aimed to elucidate the influence of the baseline value or absolute change of serum creatinine but not the overall variability of it on mortality for patients with cancer. However, the fluctuation may present a renal adaptability metric, a concept not measured by one value or absolute change. Our findings compensated for this gap and highlight the usefulness of serum creatinine fluctuation in at-risk cancer patients, allowing for risk stratification of those individuals to lower cancer mortality.
Cancer prognosis is of keen interest to cancer patients, their caregivers, and providers owing to the high mortality rates.41,42 An increasing body of research has focused on developing prognostic tools to guide patient-physician communication and decision-making. Until now, more than thirty interactive cancer prognostic tools have been developed, which incorporate cancer, demographic, and genetic characteristics, co-morbidities and therapy information, and modifiable risk factors.27,28 However, these prognostic tools largely fail to account for biomarker changes over time, such as the variability of creatinine. Our study sheds new light on the development of new cancer prognostic tools which take the variability of biomarkers into account.
Our study showed robust associations between creatinine fluctuation and poor mortality outcomes in almost all subgroups. However, what is unexpected is the negative association among patients with renal carcinoma. A potential explanation may be the small sample size, especially since the case numbers were rather low, which vary in size from seven to dozens in each group. Confounding factors including drug therapies and treatment to prevent kidney disease progression are other possible reasons. This suggests that further research is needed to fully understand the relationship between serum creatinine variability and one-year mortality in renal carcinoma patients.
The present study was not designed to explore the mechanism, but several potential explanations may explain these results. First, the variability of serum creatinine is usually ascribed to changes in GFR and reflects the capacity to adapt to nephrotoxic insults.43 Elevated change of GFR means poor ability in maintaining renal homeostasis and is independently associated with all-cause mortality.10,43 Second, creatinine generation rate is another important indicator for the variability of serum creatinine. Both the loss of muscle mass and poor nutrition can induce the change in creatinine generation rate,44 and they are also strong risk factors for cancer mortality.45 Finally, similar to AKI, a larger creatinine fluctuation frequently occurs in patients with tumors caused by chemotherapy-associated nephrotoxicity,46 which always indicates an advanced stage of cancer. Despite the mechanisms under the association between creatinine fluctuation and mortality are not well understood, we believe that this finding has important clinical implications for the identification of high-risk cancer patients.
This study has some limitations. First, we did not include any information about drug therapy during hospitalization. These pharmacologic strategies hold great promise to impact the variability of creatinine during hospitalization. However, given that all participants were patients with cancer, we performed the subgroup analysis based on cancer type, which accounted for some degrees of the confounding of treatment. Second, limited by the availability and accessibility of relevant data, the primary outcome was only one-year mortality without a longer follow-up. But one-year mortality is a commonly used endpoint in cancer research and is often considered clinically meaningful, for example, identifying patients who may require closer monitoring during hospitalization. Third, we did not define AKI based on the KDIGO AKI definition given the heterogeneity and high missing rate of urine output. A previous study conducted by Bhatraju1 et.al demonstrated that patients with different AKI sub-phenotypes defined by creatinine trajectory had different risks for mortality.32 Therefore, we performed a subgroup analysis according to the degree of creatinine variability and renal failure defined by the ICD code to address the effect of kidney status. Fourth, patients with more advanced cancer are associated with a longer hospital stay and larger measurement times of creatinine, which could induce larger creatinine fluctuation. So, the potential effects of reverse causality cannot be excluded. Fifth, it is not possible to impute causality or provide a deeper mechanistic investigation due to the observational study design of this study. Instead, our findings provided exciting novel clues awaiting now in-depth analysis in follow-up studies. Finally, only Chinese patients were involved, which may limit the generalizability of these findings to other populations.
Despite these limitations, the results of this study have the potential to inform clinical practice by providing important information on the role of creatinine variability in the care of cancer patients. Given this finding, it is important for healthcare providers to monitor creatinine levels closely in patients with malignancy to identify fluctuations and intervene early to mitigate the risk of death. Further, new strategies considering the variability of creatinine for monitoring and mitigating the risk of adverse outcomes for cancer patients could be developed. Collectively, the relationship between blood creatinine fluctuations and the risk of death in patients with malignancy is complex and requires ongoing research and monitoring.
In aggregate, the study highlights the importance of monitoring and controlling serum creatinine levels in cancer patients during hospital admission to reduce the risk of mortality. And provides important information on developing prevention strategies and triage decisions, and improves the survival of patients with cancer. Above all, our observations have important clinical implications and call for further research in this area.