Impact of kidney size on mortality in diabetic patients receiving peritoneal dialysis

Although patients with diabetes mellitus mostly present with enlarged or normal-sized kidneys throughout their life, a small proportion of patients have small kidneys. This longitudinal study enrolled 83 diabetic patients treated with peritoneal dialysis (PD) between 2015 and 2019. Patients were stratified into two groups, those with enlarged or normal (n = 67) or small (n = 16) kidneys, based on their kidney sizes before dialysis. Patients with small kidney size were not only older (76.63 ± 10.63 vs. 68.03 ± 11.26 years, P = 0.007), suffered longer duration of diabetes mellitus (272.09 ± 305.09 vs. 151.44 ± 85.31 month, P = 0.006) and predominantly female (75.0 vs. 41.8%, P = 0.017), but also had lower serum levels of creatinine (9.63 ± 2.82 vs. 11.74 ± 3.32 mg/dL, P = 0.022) and albumin (3.23 ± 0.67 vs. 3.60 ± 0.47 g/dL, P = 0.010) than patients with enlarged or normal kidney size. At the end of analysis, 14 (16.9%) patients died. Patients with small kidney size demonstrated higher all-cause (50.0 vs. 9.0%, P < 0.001) and infection-related (43.8 vs. 7.5%, P < 0.001) mortality than patients with enlarged or normal kidney size. In a multivariate-logistic-regression model, small kidney size was a powerful predictor of mortality (odds ratio 6.452, 95% confidence interval 1.220–34.482, P = 0.028). Diabetic patients with small kidney size at the beginning of PD carry a substantial risk for mortality.


Scientific Reports
| (2021) 11:8203 | https://doi.org/10.1038/s41598-021-87684-z www.nature.com/scientificreports/ Jalisco region of Mexico reported the highest incidences of treated ESKD, with rates of 493, 378, and 355 patients per million general population, respectively. Notably, diabetes is the primary cause of ESKD in 46% of incident dialysis patients in Taiwan. Nevertheless, the majority of uremic patients in Taiwan are treated with hemodialysis rather than peritoneal dialysis (PD) 7 . The rationale for this study was based on an important, but as yet unanswered, question that arose for many diabetic patients treated with PD in our hospital. The majority of diabetic patients had enlarged or normal kidneys when entering dialysis, whereas few diabetic patients were found to have small kidneys at the time of uremia. Therefore, this raises the question of what the impact of kidney size before dialysis on the outcomes of these patients is. Perhaps one of the potential clinical applications of small kidneys is to remind clinicians about diabetic patient care. For example, if small kidney size at the beginning of PD is found to be associated with a greater risk for infection-cause mortality, physicians should be alert of the possibility of occurrence of infection and consider early initiation of antimicrobial therapy in case of infectious symptoms and signs.
Diabetes mellitus is the most important cause of ESKD worldwide, but no work has been performed to compare outcomes of diabetic patients with enlarged or normal versus small kidney size, which initiated our interest in this research. Therefore, this study attempted to survey kidney size in diabetic PD patients before the commencement of dialysis and to analyze the association of kidney size with outcomes and laboratory biomarkers.

Results
This study included 83 diabetic patients receiving long-term PD at Chang Gung Memorial Hospital ( Table 1). Most of the diabetic patients had enlarged or normal kidney size (n = 67, 80.7%) at the beginning of PD, but some patients had small kidney size (n = 16,19.3%) when entering PD. The patients were aged 69.70 ± 11.57. These patients had been receiving PD for 42.82 ± 3.72 months. None of the patients were on immunosuppressive medications. Many patients suffered from hypertension (90.4%) and cardiovascular disease (41.0%). It was found that patients with small kidney size were not only older (76.63 ± 10.63 vs. 68.03 ± 11.26 years, P = 0.007) and suffered longer duration of diabetes mellitus (272.09 ± 305.09 vs. 151.44 ± 85.31 month, P = 0.006), but also more often female (75.0 vs. 41.8%, P = 0.017) than patients with enlarged or normal kidney size.
Compared with patients with enlarged or normal kidney size (Table 2), patients with small kidney size had lower serum levels of creatinine (9.63 ± 2.82 vs. 11.74 ± 3.32 mg/dL, P = 0.022) and albumin (3.23 ± 0.67 vs. 3.60 ± 0.47 g/dL, P = 0.010). Table 3 shows that there were no significant differences between the groups in terms of peritoneal transporter characteristics, dialysis adequacy, residual kidney function or cardiothoracic ratio.
At the end of analysis, 14 (16.9%) patients died (Table 4). Patients with small kidney size demonstrated a higher mortality rate than patients with enlarged or normal kidney size (50.0 vs. 9.0%, P < 0.001). Moreover, there was more infection-related mortality in the patients with small than in those with enlarged or normal kidney size (43.8 vs. 7.5%, P < 0.001). The sources of infection-related mortalities were three PD-related peritonitis, three nosocomial pneumonia, two pressure sore infection, one fulminant clostridium difficile colitis, one cellulitis, one diabetic foot wound infection and one pulmonary tuberculosis. Table 1. Baseline characteristics of peritoneal dialysis patients stratified by kidney size (n = 83). *P < 0.05; **P < 0.01.

Variable
All patients (n = 83)   Table 3. Dialysis-related data of peritoneal dialysis patients stratified by kidney size (n = 83). Residual kidney function was determined by urinary creatinine clearance using a 24-h urine collection method. Cardiothoracic ratio was measured by dividing the biggest transverse diameter of the heart silhouette by the length between the internal margins of the ribs at the level of right hemidiaphragm.  (Table 5), it was revealed that small kidney size was a significant risk factor associated with mortality (odds ratio 6.452, 95% confidence interval 1.220-34.482, P = 0.028).
Kaplan-Meier analysis revealed that patients with small kidney size had lower cumulative survival than did patients with enlarged or normal kidney size (Fig. 1, log-rank test, chi-squared = 15.614, P < 0.001).

Discussion
The analytical data from this study suggested that small kidney size at the commencement of dialysis could be a risk factor for mortality in the diabetic PD population. This finding adds new knowledge to the existing dialysis literature. There was no clear explanation for the observation. Nevertheless, patients with small kidney size were not only older (P = 0.007), suffered longer duration of diabetes mellitus (P = 0.006), but also had lower serum levels of creatinine (P = 0.022) and albumin (P = 0.010) than patients with enlarged or normal kidney size. The abovementioned variables, for example older age, longer duration of diabetes mellitus, as well as malnutrition (lower blood creatinine and albumin level) could attribute to increased mortality.
Many traditional risk factors for mortality have been described in diabetes patients receiving PD. In a 2-year nationwide cohort study, Abe et al. 8 found that glycated albumin ≥ 20.0% was associated with a decrease in survival in diabetic patients on PD. In a study of 118 diabetic PD patients, Coronel et al. 9 disclosed that age and cardiovascular comorbidity are the factors associated with mortality. In a retrospective study, Chung et al. 10 revealed that old age, female sex, the presence of cardiovascular disease or protein-energy wasting and low residual kidney function were predictors of mortality in diabetic PD patients. In a study of 2798 diabetic PD patients, Duong et al. 11 reported that poor glycemic control was associated incrementally with higher mortality. In a study of 61 diabetic PD patients, Koc et al. 12 showed that hypoparathyroidemia, hypocalcemia and hypoalbuminemia were risk factors associated with mortality. In a 5-year cohort study of 809 diabetic PD patients, Yang et al. 13 reported that older age and the presence of cardiovascular disease, hyperglycemia, anemia and hypoalbuminemia were important risk factors for mortality. In another study of 200 diabetic PD patients, Peng et al. 14 showed that increased glycated hemoglobin and decreased albumin-corrected glycated serum protein were associated with mortality in diabetic PD patients.
Patients with small kidney size were older than patients with enlarged or normal kidney size (P = 0.007). Many groups have confirmed such a positive association between decreasing kidney size and aging [15][16][17][18] . Hommos et al. 16 reported that aging could be associated with considerable changes in kidney parenchymal structure, even in the absence of age-related comorbidities. Gross pathology examination of aged kidneys found a decrease in kidney cortical volume, an increase in surface roughness, and an increase in the number and size of simple kidney cysts with aging. Histopathology examination also found an increase in nephrosclerosis (arteriosclerosis/arteriolosclerosis, global glomerulosclerosis, interstitial fibrosis, and tubular atrophy) associated with the aging process.
Patients with small kidney size were more often female than patients with enlarged or normal kidney size (P = 0.017). This positive association between small kidney size and female sex has been confirmed by many independent studies 17,18 . In a study of 665 adult volunteers, Emamina et al. 17 reported a sex difference in kidney size in adults; the median kidney lengths were 11.0 cm (women) and 11.5 (men) on the left side and 10.7 cm (women) and 11.2 cm (men) on the right side. Therefore, the female kidney was usually smaller than the male kidney. On the other hand, Piras et al. 1 revealed that female sex was associated with a 2.035-fold risk of having a large kidney. Finally, it was unclear whether the diabetic females had more history of analgesic use than diabetic males, although there were no medical records for this, and the recall history was subjected to memory bias. A questionnaire survey in Norway revealed 19 that women used both more nonprescription analgesics and more prescription analgesics than men.
The average age of patients in this study was 69.70 ± 11.57 years. This figure is close to that of the Japanese study. In Japan, patients are approximately 69 years of age at the initiation of dialysis 20 . Currently, there has been a trend of shifting from hospital-based care of older patients to home-based care. Assisted PD has been promoted as an alternative method for older patients with ESKD 21 . Older patients are susceptible to many physiological changes related to aging and problems such as anxiety, depression, dementia, visual impairment, and cognitive impairment, all of which impede self-performance of PD. Assistance from home-care nurses or assistance from a younger family member may be the solution. In a study of 128 Japanese PD patients, Sakai and Nihei 20 reported that the older group (≥ 70 years of age) did not show higher rates of technical failure, but their survival was shorter than that of the younger group (< 70 years of age). In Taiwan, most families are large families, with three generations living in the same household. Younger family members can take care of their older parents and assist them in performing PD therapy. This may explain the higher age of PD in this study. Table 4. Outcomes of peritoneal dialysis patients stratified by kidney size (n = 83). ***P < 0.001.

Variable
All patients (n = 83) Diabetic patients with enlarged or normal kidney size (n = 67) Diabetic patients with small kidney size (n = 16) P value  www.nature.com/scientificreports/ Patients with small kidney size patients suffered from lower serum levels of creatinine (P = 0.022) and albumin (P = 0.010) than patients with enlarged or normal kidney size. Albumin is the most abundant protein in human serum 22 . Albumin is produced by the liver and is an indicator of malnutrition. On the other hand, creatinine is the end product of creatine, is mainly present in skeletal muscle and is used as a surrogate measurement of muscle mass 23 . Creatinine generation is reduced in the setting of low skeletal muscle mass, such as in older patients.
Since patients with small kidney size were older and with a predominance of females, it was not surprising that these patients had lower serum levels of creatinine and albumin than the enlarged or normal kidney size group.
Patients with small kidney size demonstrated higher all-cause (P < 0.001) and infection-related (P < 0.001) mortality than patients with enlarged or normal kidney size. Diabetics generally have an increased propensity to develop infections. The increased occurrence of infections in diabetic patients is caused by the hyperglycemic milieu, which supports immune dysfunction (such as the impairment of neutrophil function, the suppression of antioxidant system, and humoral immunity), angiopathy, neuropathy, a decrease in the antibacterial activity of urine, gastrointestinal and urinary dysmotility, and an increased need for medical interventions 24 . Diabetes impairs antibacterial defenses and increases the risk of infection 25 . Furthermore, a high concentration of glucose degradation products in PD solutions could accelerate leucocyte apoptosis and damage peritoneal antibacterial defense 26 . Previous studies 27,28 have reported diabetes mellitus as a risk factor for PD-associated peritonitis. According to a large retrospective cohort study in China 29 , peritonitis was always associated with a higher risk of mortality in PD patients, and its influence on mortality was more significant in patients with longer PD durations. The assessment of the immune defense system of the diabetic patients with small kidney size will need further investigation.
The limitations of this study included small sample size and short follow up duration. In addition, the relationship between small kidneys and mortality was simply a correlation, rather than causal relationship. Since this study involves retrospective review of existing data, it is impossible to make a causal relationship conclusion. The pathophysiological mechanism by which small kidneys confer increased infection risk cannot be answered by this study. Although patients with small kidney size were older and suffered longer duration of diabetes mellitus than patients with enlarged or normal kidney size, there were no difference in blood sugar control between both groups, which could confer increased infection risk. This study is also limited by lacking protocol kidney biopsy. Since approximately half of cases of ESKD in Taiwan are due to diabetic nephropathy, it is possible that half of this cohort may have had other kidney diseases. All patients in this study are diabetes, but the etiologies of renal failure are varied. Since kidney biopsy remains an invasive procedure, it is usually not considered a routine in patients with diabetes mellitus unless suspicion of non-diabetic renal disease. As shown in Table 1, only one patient received kidney biopsy and the pathology report revealed co-existing glomerulonephritis. Further prospective studies are warranted to confirm this finding.

Conclusion
In summary, small kidney size at the beginning of dialysis carries a substantial risk for mortality in diabetic PD patients. These patients demonstrated higher all-cause and infection-related mortality than patients with enlarged or normal kidney size.

Methods
Inclusion and exclusion criteria. All diabetic patients aged 18 years and above receiving chronic PD at Chang Gung Memorial Hospital between 2015 and 2019 were included in this study. Patients who had been receiving PD for less than 6 months; those who had been hospitalized or operated on or who had received a kidney transplant in the preceding 3 months; and those with cancer were excluded from the study (Fig. 2).

Groups.
Patients who met the inclusion criteria were divided into two groups according to their kidney size when entering PD. All patients received ultrasonographic evaluation of kidney size and echogenicity, and a small kidney was defined as when the kidney length was less than 9.0 cm 4 . PD prescription. PD prescriptions were based on the peritoneal membrane characteristics determined by the peritoneal equilibration test 30 . Intermittent therapy was prescribed to patients with high membrane transport and continuous therapy and to those with average or low membrane transport. Low-calcium (1.5 or 1.25 mmol/L), icodextrin-based (7.5 g/dL) or standard dialysates containing glucose (sodium, 135 mmol/L; lactate, 35 mmol/L; calcium, 1.75 mmol/L) were used according to the patients' peritoneal transport characteristics and serum calcium levels to maintain adequate ultrafiltration and enlarged or normal calcium levels. Dialysis prescription aimed at obtaining a total Kt/V of at least 1.8 per week.
Laboratory analysis. The data represented the last laboratory values prior to the patients being started on PD. All laboratory values, including blood cell count, biochemical data, dialysate/plasma creatinine ratio, peritoneal transport characteristics, weekly creatinine clearance and weekly Kt/V urea, were surveyed by automated and standardized methods. All blood samples were collected in the morning after at least 10 h of fasting. Serum levels of calcium, phosphate and intact parathyroid hormone were also surveyed, and the corrected serum calcium level was calculated as calcium (mg/dL) = [0.8 (4.0 − albumin[g/dL])]. All other markers were surveyed via standard laboratory methods using an automatic analyzer.
Statistical analysis. The continuous variables are presented as the means ± the standard deviations for the numbers of observations, whereas the categorical variables are presented as numbers (percentages) 31 . For comparison between two groups, Student's t-test was used for quantitative variables, whereas the chi-squared or Fisher's exact test was used for categorical variables. Survival data were analyzed with the Kaplan-Meier method and tested for significance using the log-rank test. A univariate binary logistic regression analysis was performed to compare the frequency of potential risk factors associated with mortality. To control for confounders, a stepwise backward multivariate binary logistic regression analysis was performed to analyze the variables that were significant on univariate analysis. The criterion for significance to reject the null hypothesis was a 95%