This retrospective cohort study was conducted at Seoul National University Hospital, which is a tertiary referral hospital in Seoul, South Korea. We used data from electronic medical records and the cancer registry system. In Korea, The Ministry of Health and Welfare generated a nationwide registry called the Korea Central Cancer Registry (KCCR), beginning in 1980, to register cancer patients and provide medical insurance benefits to them. All patients diagnosed with any type of cancer in Korea were registered on the KCCR based on the International Classification of Diseases 10th revision (ICD-10) [14].

We included patients who were diagnosed with cancer between January 1, 2004, and December 31, 2013 based on the KCCR registration in Seoul National University Hospital. From a total of 106,004 patients who were registered on the KCRR, we excluded patients having the following exclusion criteria after electronic medical record review: (1) aged less than 18 years, (2) registered with more than two types of cancer, (3) no measured serum creatinine (sCr) or checked only once, (4) initial estimated glomerular filtration rate (eGFR) < 15 mL/min/1.73 m² or undergoing dialysis before cancer diagnosis, and (5) no accurate data for death and follow-up duration, as previously reported [14]. Therefore, 67,986 patients were eligible for this study, and among them, 5,855 were patients diagnosed with respiratory tract cancer. We selected the final study subjects who had been diagnosed with lung cancer and by setting additional exclusion criteria: (1) no accurate data for demographic information, laboratory findings, and treatment history, (2) development of AKI within 2 weeks before death, and (3) initial sCr < 0.6 mg/dL.

We reviewed and collected data from the electronic medical records. Demographic characteristics included the age, sex, body mass index (BMI), and smoking history. Systolic and diastolic blood pressure was measured at the time of the first hospital admission. The data regarding a history of diabetes mellitus (DM) was also gathered. Initial laboratory findings including initial sCr, anemia, and serum level of calcium, phosphorus, sodium, potassium, chloride, total CO₂, uric acid, albumin, and alkaline phosphatase (ALP) were collected at the time of cancer diagnosis. The eGFR was calculated using the CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) formula [16].

Like other solid organ cancers, lung cancer consists of various histological subtypes and the cancer stage determines the treatment strategy. Both histological subtypes and the cancer stage could affect AKI occurrence. Therefore, we investigated cancer-specific factors, such as the pathological subtype, cancer stage, and treatment modality. Pathological subtypes were divided into four groups: adenocarcinoma, squamous cell carcinoma, small cell carcinoma, and others. Most of the “others” group included non-small cell carcinoma, which was difficult to classify into more detailed pathological subtypes such as adenocarcinoma or squamous cell carcinoma. Carcinoid, neuroendocrine tumor, and sarcoma were also included in the “others” group. We used the cancer staging system provided by National Institutes of Health in the United States, which is named the Surveillance, Epidemiology, and End Results (SEER) stage [17]. The SEER stage is the most basic way of categorizing how far a cancer has spread from its point of origin. It consists of nine codes: in situ, localized only, regional by direct extension only, regional lymph nodes involved only, regional by both direct extension and lymph node involvement, regional not otherwise specified, distant site(s)/node(s) involved, and unknown if extension or metastasis (un-staged, unknown, or unspecified) [17]. Treatment modalities were divided into four groups: surgery only, surgery and chemotherapy, chemotherapy only, and neither surgery nor chemotherapy, respectively. Radiation therapy was excluded from our analysis because it was thought that radiation had little effect on the kidney function [18]. We investigated whether the chemotherapeutic agents known to have nephrotoxicity including cisplatin, carboplatin, ifosfamide, gemcitabine, and pemetrexed were administered during the study period [19]. We reviewed the guidelines and chose the drugs based on the chemotherapy regimen of lung cancer used in our institution [20]. A count of radiocontrast-enhanced computed tomography per year was also explored to consider possible contrast-induced AKI.

The definition of AKI was applied in the same way as in our previous study [14]. Initial sCr was defined as the first measured sCr within 2 months before or after the KCCR registration date. Considering possible fluctuating kidney function during cancer diagnosis and treatment in lung cancer patients, we defined the baseline sCr as the minimum value of sCr from the current time to the previous 3 weeks by shifting the reference point every 3 weeks based on the KCCR registration date for a more precise analysis.

We defined AKI and the stage of AKI based on the Kidney Disease: Improving Global Outcomes (KDIGO) AKI guidelines [21] using sCr levels. Urine output criteria could not be considered due to lack of complete urine output data. Patients were classified according to the presence and its severity of AKI into three groups as no AKI, AKI stage 1, and AKI stage 2/3.

The primary outcome was to identify the incidence of AKI and its impact on all-cause mortality in patients with lung cancer. In the mortality analysis, we analyzed the patients who survived for > 1-year after lung cancer diagnosis because we intended to determine the direct impact of AKI on mortality. If a patient died within 1 year of the cancer diagnosis, even if they had AKI, we could not presume that the incident AKI affected the death. Subgroup analysis was performed according to age, sex, pathologic subtype, and SEER stage to determine the effect of advanced AKI on mortality.

The secondary outcome was long-term kidney outcomes. The last eGFR was calculated as the mean value of eGFRs based on the last three sCr measurements. The average of the last three eGFRs was categorized in five groups: (1) 60 or more, (2) 45 or more and less than 60, (3) 30 or more and less than 45, (4) 15 or more and less than 30, and (5) less than 15. In the same way, the initial eGFR was also classified, and the change of the stage was examined. We defined a reduction of the eGFR to less than 60 mL/min/1.73 m² as new-onset chronic kidney disease (CKD). In addition, we explored cases of progression to end-stage kidney disease (ESKD) requiring maintenance renal replacement therapy such as hemodialysis and peritoneal dialysis. To assess long-term outcomes, we divided the proportions of early and last eGFR into quartiles and looked at their distribution.

We presented continuous variables as mean±standard deviation, and categorical variables as percentage. We used the student t test, one-way analysis of variance (ANOVA), and chi-square test to investigate the differences among the no AKI group, AKI stage 1 group, and AKI stage 2/3 group. A multivariable logistic model was developed to identify the independent risk factors of AKI using the odds ratio (OR) and 95% confidence interval (CI). Cox regression analyses were used to calculate the hazard ratio (HR) and 95% CI for the risk of all-cause mortality. The Kaplan-Meier survival curve was used to compare the cumulative mortality among the no AKI group, AKI stage 1 group, and AKI stage 2/3 group with the p-value calculated from the log-rank method. The SPSS ver. 25 (IBM Corp., Armonk, NY) and R software ver. 3.6.1 (R Software, R Foundation for Statistical Computing, Vienna, Austria) were used to perform the statistical analysis. A p-value < 0.05 was considered statistically significant.

A total of 3,202 patients were enrolled in this study (S1 Fig.), and the mean follow-up duration was 2.6±2.2 years. The mean age was 63.8±10.3 years, and 31.4% of the patients were female as shown in Table 1. Patients with DM and CKD with an initial eGFR < 60 mL/min/1.73 m² were 13.8% and 11%, respectively. A total of 3,061 patients (95.6%) underwent radiocontrast exposure by computed tomography more than once; the mean annual number of radiocontrast exposures in these patients was 4.3±3.9. Overall, the most common pathological subtype was adenocarcinoma (56%), followed by squamous cell carcinoma (23.1%), small cell carcinoma (10.6%), and others (10.1%). We found that 35.9% of the patients were at the “regional” SEER stage at the initial diagnosis, and “distant” (27.6%), “localized or in situ” (23.3%), and “unknown” (13.1%) SEER stages following that. Chemotherapeutic agents were used in 63.5% of the enrolled patients (Table 2).

According to the severity of AKI, three groups were categorized as the no AKI group, AKI stage 1 group, and AKI stage 2/3 group, and accounted for 55.7%, 42.1%, and 2.1%, respectively. After diagnosis of lung cancer, it took 0.64±1.10 years to develop first AKI. The more severe AKI was, the older the patients were. Males were predominant in all AKI groups. The severity of AKI was correlated with higher systolic blood pressure, increased annual radiocontrast exposure, a higher proportion of DM and anemia, and higher proportion of initial eGFR < 60 mL/min/1.73 m². Among the pathological subtypes, small cell carcinoma was more significantly associated with AKI than other subtypes. Patients receiving chemotherapy tend to develop AKI, although, unexpectedly, the use of nephrotoxic chemotherapeutic agents including cisplatin, carboplatin, ifosfamide, gemcitabine, and pemetrexed seemed not to be associated with AKI risk elevation.

A total of 68 patients experienced severe AKI (stage 2/3); their clinical characteristics are shown in Table 3. The most common cause of severe AKI was sepsis (44.1%), followed by ischemia (11.8%), nephrotoxic drugs (8.8%), cardio-renal syndrome (2.9%), and others (10.3%). The majority of these patients recovered from AKI (92.6%); however, the remaining 4.4% of patients experiencing severe AKI eventually progressed to kidney failure requiring maintenance renal replacement therapy. Ten patients (14.7%) died after the development of AKI. The mean duration from the development of AKI to death was 48.4±50.6 days.

We describe the clinical risk factors of AKI occurrence in lung cancer patients during their cancer diagnosis and treatment in S2 Table. Among the clinical characteristics, older age, male sex, higher systolic blood pressure, presence of DM, anemia, and hyponatremia, lower initial eGFR < 60 mL/min/1.73 m², and lower serum albumin were independent risk factors of AKI development after lung cancer diagnosis. Increased annual radiocontrast exposure and receiving chemotherapy were also correlated with an elevated risk of incident AKI, although SEER stage was not.

Among the pathological subtypes, squamous cell carcinoma (adjusted OR, 1.29; 95% CI, 1.047 to 1.590; p=0.017) and small cell carcinoma (adjusted OR, 1.57; 95% CI, 1.188 to 2.067; p=0.001) were associated with AKI occurrence. When comparing baseline characteristics according to the pathological subtypes, the proportion of patients with risk factors of AKI in small cell carcinoma and squamous cell carcinoma was statistically significantly higher (S3 Table).

To elucidate whether AKI affected lung cancer patient survival, we analyzed the effect of AKI on mortality only among patients who survived > 1 year after cancer diagnosis. During the study period, 1,251 patients (53.7%) were died, and overall 5-year survival rate was 46.9%. The Kaplan-Meier curve showed that more severe AKI was related to a lower survival rate (Fig. 1). As shown in Table 4, the incident AKI elevated all-cause mortality in a dose-responsive manner (AKI stage 1: adjusted HR, 1.15; 95% CI, 1.022 to 1.292; p=0.020 and AKI stage 2/3: adjusted HR, 1.73; 95% CI, 1.194 to 2.505; p=0.004), even after the adjustment of age, sex, BMI, smoking history, initial eGFR < 60 mL/min/1.73 m², presence of anemia and hyponatremia, and pathological subtypes.

Subgroup analyses showed that even mild AKI (stage 1) was associated with increased mortality, especially in women and adenocarcinoma cancer type. Severe AKI (stage 2/3) was associated with elevated mortality in every subgroup except women, small cell carcinoma, and distant and unknown SEER stage shown in S4 Fig.

The effect of AKI on long-term kidney outcomes was shown in Table 5 and S5 Table. The incidence rate of the new-onset CKD was 7.1% and it was associated with AKI experiences with dose-responsive manner. When comparing the CKD stage of the last eGFR with the initial eGFR, most lung cancer patients experienced worsening kidney function. Particularly, patients suffered from more severe AKI tended to face more worsening CKD stages. Among the enrolled patients, we found that 21 patients (0.7%) progressed to ESKD requiring maintenance renal replacement therapy after lung cancer diagnosis during the study period, and all of them experienced prior AKI (Table 5). Their median time to ESKD progression was 2.16±2.71 years. Furthermore, the ratio of initial eGFR and the average of the last three eGFRs (last three eGFRs/initial eGFR) was lower in patients with more advanced AKI (stage 2/3) (S5 Table).