bs_bs_banner
Nephrology 20 (2015) 936–944
Original Article
Validation of chronic kidney disease risk categorization system in Chinese patients with kidney disease: A cohort study QINGYAN LIU,1 JICHENG LV,2,3,4 HAIXIA LI,1 LILI JIAO,1 HONGYUN YANG,1 YINAN SONG1 and GUOBIN XU1 1 Department of Clinical Laboratory, 2Division of Nephrology and Institute of Nephrology, Peking University First Hospital, 3Key Laboratory of Renal Disease, Ministry of Health of China, and 4Key Laboratory of Chronic Kidney Disease Prevention and Treatment (Peking University), Ministry of Education, Beijing, China
KEY WORDS: chronic kidney disease, glomerular filtration rate, prognosis, proteinuria. Correspondence Prof Guobin Xu, No. 8, Xishiku Street, Xicheng District, Beijing 100034, China. Email: [email protected] Accepted for publication 27 May 2015. Accepted manuscript online 1 June 2015. doi:10.1111/nep.12528 Conflict of interest: The authors declare no conflict of interest.
SUMMARY AT A GLANCE This paper provides a validation of the KDIGO risk stratification in a Chinese population living in China, based on a combination of eGFR and proteinuria.
ABSTRACT Aim: To validate the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines risk stratification system based on the combination of estimated glomerular filtration rate (eGFR) and proteinuria. Methods: This was a cohort study. A total of 1219 study population were recruited. Estimated GFR and proteinuria measured by using 24 h urine protein excretion rate (PER) were predictors. Adverse outcomes included all-cause mortality (ACM) and end-stage renal disease (ESRD). Follow-up was done by regular visit, telephone interview and electronic medical records. Results: Over a median follow-up of 4.6 years, 153 (12.6%) and 43 (3.5%) patients experienced ESRD and ACM, respectively. On multivariable analysis, the adjusted hazard ratio for ESRD and ACM (compared with patients with eGFR > 60 mL/min per 1.73 m2) was of 29.8 and 3.6 for those with eGFR of 15–29 mL/min per 1.73 m2, respectively. The adjusted hazard ratio for ESRD and ACM (compared with patients with PER < 150 mg/24h) was of 15.9 and 3.9 for those with PER > 500 mg/24h. Higher KDIGO guidelines risk categories (indicating lower eGFR or higher proteinuria) were associated with a graded increase in the risk for the ESRD (P < 0.001) and ACM (P < 0.001). Reclassification of KDIGO guidelines risk categories yielded net reclassification improvements for those with ESRD or ACM event (NRIevents) of 33.3% or 30.2%. Conclusion: Lower eGFR and higher proteinuria are risk factors for ESRD and ACM in Chinese patients. The KDIGO guidelines risk categorization system assigned patients who went on to have the event to more appropriate CKD risk categories.
The Kidney Disease Outcomes Quality Initiative (KDOQI) guidelines1 risk categorization system, which relies almost exclusively on estimated glomerular filtration rate (eGFR), does not incorporate information about how the presence and severity of proteinuria might affect prognosis for important clinical outcomes in each chronic kidney disease (CKD) risk category.2–5 The vast body of evidence provided by the CKD Prognosis Consortium6 demonstrated that both eGFR and albuminuria were associated with adverse kidney outcomes, and all-cause and cardiovascular mortality, independent of traditional cardiovascular risk factors and independent of each other and despite inclusion of diverse study 936
populations.7–15 The Kidney Disease: Improving Global Outcomes (KDIGO) guidelines,16 therefore recommended CKD staging to be based on both eGFR and proteinuria and proposed four risk categories according to the combination of eGFR categories with proteinuria categories. However, data for validation of the KDIGO guidelines risk categorization system in Chinese patients is limited and the prognostic implication of stratifying patients by combining eGFR categories with 24 h urine protein excretion rate (PER) categories has not yet been investigated among CKD patients with a broad spectrum of underlying aetiologies. We validated that the KDIGO guidelines risk categorization system more © 2015 Asian Pacific Society of Nephrology
Validation of CKD risk categorization system
accurately predicts risk for all-cause mortality (ACM) and end-stage renal disease (ESRD) than the KDOQI guidelines risk categorization system in this study.
METHODS Study population A total of 3043 patients with laboratory tests of 24 h endogenous creatinine clearance rate (Ccr) and 24 h urine PER on the same day between May 2009 with May 2010 in a tertiary nephrology centre were continuously enrolled. After we excluded 155 patients with previous ESRD or an initial eGFR less than 15 mL/min per 1.73 m2 (on the assumption that they would reach the end of follow-up), 251 patients who were younger than 18 years and 214 pregnant patients, and 1058 patients without contact information, a total of 1365 patients entered the follow-up. Because of unknown follow-up status, 146 patients were missed. Data were available for 1219 patients with kidney disease. Flow diagram of the study population is shown in Figure 1. Data on demographics, history of diabetes mellitus, hypertension, co-morbidities and current drug intake were recorded by a standardized questionnaire. Laboratory values and contact information were determined from the administrative data files in the laboratory information system and hospital information system of Peking University First Hospital. Diabetes mellitus, hypertension, or other comorbid conditions was identified by American Association of Clinical Endocrinologists medical guidelines for clinical practice for the management of diabetes mellitus,17 2007 ESH-ESC Practice Guidelines for the Management of Arterial Hypertension,18 or KDIGO Clinical Practice Guidelines for Glomerulonephritis.19 A specialist in nephrology confirmed the aetiology of CKD by medical history, clinical manifestation, laboratory examination, and kidney biopsy. This study was approved by the Medical Ethics Committee of Peking University First Hospital. All patients provided oral or written consent to their participation in the follow-up.
Assessment of kidney function and proteinuria Estimated GFR for each patient was calculated by using the Chronic Kidney Disease-Epidemiology Collaboration (CKD-EPI) two-level race creatinine equation.20 Serum creatinine was measured in blood samples after an overnight fast, using the Jaffé method (CV ≤ 5%), which can be traceable to an isotope dilution mass spectrometry reference. For patients with multiple creatinine measurements, the mean of all measures taken within 4 months of the first measurement in the follow-up period was used (on the assumption that abnormalities of kidney function would present for >3 months), with the date of the last measurement in that 4-month period used as the baseline date. Proteinuria was measured on the basis of 24 h urine measurements by using PER, PCR and urine dipstick testing, respectively. PER was the urinary total protein concentration multiplied by volumes of 24 h urine. Urinary total protein concentration was measured by pyrogallol red colorimetric method. PCR was the urinary total protein concentration divided by urine creatinine concentration in 24 h urine. Urine creatinine concentration was determined using the Jaffé method, and then converted the concentration unit of mmol/L to g/L to calculate PCR. Urinary total protein concentration (≤100 mg/L, CV < 20%; >100 mg/L, CV < 7%) and urinary creatinine concentration (CV ≤ 5%) were measured on Hitachi 7180 Biochemical analysis system. Urine dipstick testing was measured by bromophenol blue assay. PER, PCR and urine dipstick measurements tested on the same day as Ccr measurement were used to establish baseline proteinuria. For patients with multiple PER, PCR and urine dipstick measurements, the median (using ordinal numbers for dipstick protein categories)21 of all measures taken within 4 months of the first measurement in the follow-up period was used (on the assumption that kidney damage would present for >3 months).
CKD risk category Baseline eGFR for each patient was categorized as 60 or greater, 45 to 59, 30 to 44, and 15 to 29 mL/min per 1.73 m2. Proteinuria among patients was classified as normal (PER < 150 mg/24h, PCR < 150 mg/g or urine dipstick negative), mild (PER 150 to 500 mg/24h, PCR 150 to 500 mg/g or urine dipstick trace or 1+), or heavy (PER > 500 mg/24h, PCR > 500 mg/g or urine dipstick ≥2+).16,22 Each patient was placed into a proteinuria category based on PER, PCR and dipstick results, respectively. Similarly, each patient was allocated a CKD risk category for each combination of eGFR category and proteinuria categories allocated by PER, PCR and dipstick results, respectively. CKD risk categories (0, 1, 2 and 3) were in accordance with KDOQI guidelines1 and 2012 KDIGO guidelines.16 Risk category 0, 1, 2 and 3 reflect the low, medium, high and very high risk of progression, respectively.
Adverse outcome assessment
Fig. 1 Flow diagram of the study population. © 2015 Asian Pacific Society of Nephrology
All patients were invited to regularly visit the outpatient department, and information on laboratory values was obtained from electronic medical records. Patients who did not attend regular visits were contacted via telephone. Trained research personnel making the telephone calls were blinded to patients’ baseline clinical status. Patients were asked standardized follow-up questions at the end of 937
Q Liu et al.
follow-up. Outcomes data collected included ACM and ESRD. ACM was defined as death from any cause that was confirmed by either a death certificate from the local citizen registry or the electronic medical record of Peking University First Hospital. When no official documentation was available, case fatality was decided if death was reported by different proxies. ESRD was defined as registration for long-term dialysis or renal transplantation. Cases of ESRD were cross-checked with the treating hospitals to ensure the accuracy of diagnosis. Date of ESRD initiation or death was determined from telephone interview and electronic medical records. Patients who experienced ESRD and then died were considered to have experienced both adverse outcomes, with the date for each event used to calculate follow-up time. All outcomes were assessed from the baseline date until study end (31 August 2014).
Statistical analysis We validated the KDIGO guidelines risk categorization system by using PER, PCR and dipstick cohort. PER and PCR were converted to a log scale, and then compared with a scatterplot, using both Lowess (locally weighted scatterplot smoothing) and Deming methods to fit the regression. Spearman correlation coefficients were calculated to assess the correlation between PER and PCR in the study population. PER, PCR or urine dipstick result was used to cross-tabulating together with baseline eGFR, respectively. Rates of adverse outcomes (ESRD and ACM) were calculated by Poisson regression for each of 12 groups defined by the four eGFR and three proteinuria categories, with output expressed per 1000 person-years. The association between eGFR or proteinuria and adverse outcomes was analyzed by the Cox proportional hazard model. Four CKD categories ranging from 0 (no CKD) to 3 (most severe CKD) were formed by KDIGO guidelines risk categorization system. Incidence rates of ESRD and ACM in four CKD categories were calculated as cumulative incidence and compared using the rate ratio.23 The reclassification accuracy of participants with and without adverse outcomes using KDIGO versus KDOQI guidelines risk categorization system was quantified using the net reclassification improvement (NRI),24 which was defined as the proportions of ‘clinically correct’ reclassification minus the proportions of ‘clinically incorrect’ reclassification. Specifically, the ‘clinically correct’ reclassification was the reclassification of people with adverse outcomes to higher versus lower categories and for those without adverse outcomes to lower versus higher categories. The consistency of these findings was also assessed in subgroups stratified by age ( 60 mL/min per 1.73 m2) was of 29.8 (95% CI: 16.1–55.1) and 3.6 (95% CI: 1.3–10.1) for those with eGFR of 15–29 mL/min per 1.73 m2, respectively (Table 2). Likewise, higher proteinuria was associated with higher rate of ESRD and ACM overall and within every estimated GFR category. The adjusted hazard ratio for ESRD and ACM (compared with patients with PER < 150 mg/24 h) was of 15.9 (95% CI: 2.2–116.8) and 3.9 (95% CI: 1.1–14.7) for those with PER > 500 mg/24 h, respectively (Table 3). Results were similar in the analysis of PCR (Fig. 3A2,B2) and dipstick cohort (Fig. 3A3,B31).
Cumulative incidences of ESRD and ACM by CKD category In the PER cohort, the cumulative incidences of ESRD in CKD category 0–3 were 0, 1.6, 10.5 and 116.4 per 1000 person-years, respectively and the cumulative incidences of ACM in CKD category 0–3 were 3.4, 4.0, 4.5 and 23.8 per 1000 person-years, respectively. In the PCR cohort, the cumulative incidences of ESRD in CKD category 0–3 were 0, 1.5, 12.4 and 121.5 per 1000 person-years, respectively, and the cumulative incidences of ACM in CKD category 0–3 were 3.0, 3.8, 5.3 and 24.8 per 1000 person-years, respectively. In the dipstick cohort, the cumulative incidences of ESRD in CKD category 0–3 were 0, 4.7, 28.5, 144.2 per 1000 personyears, respectively and the cumulative incidences of ACM in CKD category 0–3 were 3.2, 4.7, 4.3 and 30.5 per 1000 © 2015 Asian Pacific Society of Nephrology
Details for reclassification accuracy of the KDIGO risk classification system for ESRD are listed in Table 4. In the PER cohort, Of 153 patients who suffered ESRD events during the follow-up, one (0.7%) patient was incorrectly reclassified to a lower risk category and 53 (34%) patients were correctly reclassified to a higher risk category using the KDIGO guidelines risk classification system, yielding an NRIevents of 33.3%. In contrast, of 1066 patients who did not experience ESRD events, 47 (4.4%) patients were correctly reclassified to a lower risk category and 418 (39.2%) patients were incorrectly reclassified to a higher risk category, resulting in a NRInon-events of −34.8%. Similarly, the NRIevents and NRInon-events in the PCR and dipstick cohorts were 33.3% and −28.7%, 19.0% and −7.5%, respectively. Details for reclassification accuracy of the KDIGO guidelines risk classification system for ACM are listed in Table 5. In the PER cohort, of 43 patients who suffered ACM events during the follow-up, one (2.3%) patient was incorrectly reclassified to a lower risk category and 14 (32.6%) patients were correctly reclassified to a higher risk category using the KDIGO risk classification system, yielding an NRIevents of 30.2%. In contrast, of 1176 patients who did not experience ACM events, 47 (4.0%) patients were correctly reclassified to a lower risk category and 456 (38.8%) patients were incorrectly reclassified to a higher risk category, resulting in a NRInon-events of −34.8%. Similarly, the NRIevents and NRInon-events in PCR and dipstick cohort were 30.2% and −29.3%, 14.0% and −8.8%, respectively.
DISCUSSION This cohort study is the first to validate the KDIGO guidelines risk categorization system in Chinese CKD patients with a broad spectrum of underlying aetiologies and prove the performance of 24 h urine PER, PCR and dipstick urinalysis in predicting risk of ESRD and ACM. Overall, our results showed that decreased eGFR as calculated by CKD-EPI equation and more proteinuria as measured by 24 h urine PER, PCR, and dipstick urinalysis were associated independently with increased risk of ESRD and ACM (Fig. 3, Tables 2,3). However, the predictive strength of PCR and PER appeared to be stronger than that of the dipstick urinalysis (Tables 4,5). Moreover, the present study also showed that KDIGO guidelines risk categorization system appeared to be more appropriate than KDOQI guidelines risk categorization system in Chinese patients with kidney disease according to their risk of ESRD and ACM. KDIGO guidelines risk 939
Q Liu et al.
Fig. 3 Unadjusted incidence rates (per 1000 person-years) for (A) end-stage renal disease (ESRD) (B) all-cause mortality (ACM) in the protein excretion rate (PER), PCR and dipstick cohort. (A1) (B1): PER cohort; (A2) (B2): protein-creatinine ratio (PCR) cohort; (A3) (B3): dipstick cohort.
Table 2 Adjusted† hazard ratio (95% confidence interval) for end-stage renal disease and all-cause mortality, by proteinuria glomerular filtration rate (GFR0 category eGFR (mL/min per 1.73 m2)
ESRD ACM
≥60
45–59
30–44
15–29
Ref. Ref.
3.6 (1.7, 7.5) P = 0.001 0.8 (0.2, 2.5) P = 0.671
6.6 (3.2, 13.6) P < 0.001 0.7 (0.2, 2.9) P = 0.652
29.8 (16.1, 55.1) P < 0.001 3.6 (1.3, 10.1) P = 0.016
†Adjusted for age, sex, body mass index (BMI), diabetes mellitus, hypertension and proteinuria. ACM, all-cause mortality; ESRD, end-stage renal disease.
940
© 2015 Asian Pacific Society of Nephrology
Validation of CKD risk categorization system
Table 3 Adjusted† hazard ratio (95% confidence interval) for end-stage renal disease (ESRD) and all-cause mortality (ACM), by proteinuria category Proteinuria
PER cohort ESRD ACM PCR cohort ESRD ACM Dipstick cohort ESRD ACM
Normal
Mild
Heavy
Ref. Ref.
5.1 (0.6, 40.5) P = 0.124 1.6 (0.3, 7.2) P = 0.562
15.9 (2.2, 116.8) P = 0.007 3.9 (1.1, 14.7) P = 0.042
Ref. Ref.
7.0 (0.9, 55.0) P = 0.064 3.2 (0.8, 13.1) P = 0.110
19.6 (2.7, 144.1) P = 0.003 3.6 (0.9, 13.5) P = 0.043
Ref. Ref.
5.7 (1.4, 23.9) P = 0.016 1.8 (0.6, 5.9) P = 0.310
13.7 (3.1, 60.1) P = 0.001 4.3 (1.0, 19.3) P = 0.035
†Adjusted for age, sex, body mass index (BMI), diabetes mellitus, hypertension and estimated glomerular filtration rate (eGFR).
Table 4 Reclassification accuracy of the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines risk classification system for end-stage renal disease, by cohort and subgroup Cohort and subgroup
PER cohort Total population Age < 60 years Age ≥ 60 years Men Women Referral at category 3† PCR cohort Total population Age < 60 years Age ≥ 60 years Men Women Referral at category 3† Dipstick cohort Total population Age < 60 years Age ≥ 60 years Men Women Referral at category 3†
Patients (Events/No Events), n
Reclassification Accuracy Events, n (% with events)
No Events, n (% without events)
Incorrect
Correct
Net
Incorrect
Correct
Net
1219 (153/1066) 882 (97/785) 337 (56/281) 634 (94/540) 585 (59/526) 1219 (153/1066)
1 (0.7) 0 (0) 1 (1.8) 0 (0) 1 (1.7) 0 (0)
52 (34.0) 40 (41.2) 12 (21.4) 34 (36.2) 18 (30.5) 40 (26.1)
51 (33.3) 40 (41.2) 11 (19.6) 34 (36.2) 17 (28.8) 40 (26.1)
418 (39.2) 330 (42.0) 88 (31.3) 232 (43.0) 186 (35.4) 159 (14.9)
47 (4.4) 15 (1.9) 32 (11.4) 31 (5.7) 16 (3.0) 0 (0)
−371 (−34.8) −315 (−40.1) −56 (−19.9) −201 (−37.2) −170 (−32.3) −159 (−14.9)
1219 (153/1066) 882 (97/785) 337 (56/281) 634 (94/540) 585 (59/526) 1219 (153/1066)
1 (0.7) 0 (0) 1 (1.8) 0 (0) 1 (1.7) 0 (0)
52 (34.0) 40 (41.2) 12 (21.4) 34 (36.2) 18 (30.5) 40 (26.1)
51 (33.3) 40 (41.2) 11 (19.6) 34 (36.2) 17 (28.8) 40 (26.1)
358 (33.6) 276 (35.2) 82 (29.2) 175 (32.4) 183 (34.8) 148 (13.9)
52 (4.9) 19 (2.4) 33 (11.7) 36 (6.7) 16 (3.0) 0 (0)
−306 (−28.7) −257 (−32.7) −49 (−17.4) −139 (−25.7) −167 (−31.8) −148 (−13.9)
1219 (153/1066) 882 (97/785) 337 (56/281) 634 (94/540) 585 (59/526) 1219 (153/1066)
2 (1.3) 0 (0) 2 (3.6) 1 (1.1) 1 (1.7) 0 (0)
31 (20.3) 25 (25.8) 6 (10.7) 22 (23.4) 9 (15.3) 26 (17.0)
29 (19.0) 25 (25.8) 4 (7.1) 21 (22.3) 8 (13.6) 26 (17.0)
130 (12.2) 83 (10.6) 47 (16.7) 77 (14.3) 53 (10.1) 96 (9.0)
50 (4.7) 18 (2.3) 32 (11.4) 32 (5.9) 18 (3.4) 0 (0)
−80 (−7.5) −65 (−8.3) −15 (−5.3) −45 (−8.3) −35 (−6.7) −96 (−9.0)
†Reclassification is defined as movement from referral to nonreferral (risk category 3 in the old system to risk category 1 or 2 in the new system) or vice versa.
categorization system reclassified more of the patients who went on to have an outcome event to more advanced CKD risk categories, and correct reclassification was more likely for the ESRD than for ACM (Tables 4,5). The KDIGO guidelines risk categorization system has important clinical implications for identifying additional patients who are in need of closer follow-up and more aggressive treatment. The KDIGO guidelines risk categorization system was developed predominantly on the basis of a white population.6 However, whether the KDIGO guidelines risk categorization system more accurately predicts risks for ACM and © 2015 Asian Pacific Society of Nephrology
ESRD than the KDOQI guidelines risk categorization system in Chinese patients with kidney disease has not been tested before. Our study complements and extends the previous studies25,26 in Chinese patients with kidney disease, and supports recent recommendations defining chronic kidney disease and stratifying subsequent risks based on both decreased GFR and proteinuria. This study has several advantages. First, our analysis was performed in Chinese patients with CKD, which was predominantly secondary to glomerular disease (45%). The overall incident of ESRD is higher than ACM. It is different 941
Q Liu et al.
Fig. 4 Cumulative incidences of (A) end-stage renal disease (ESRD)and (B) all-cause mortality (ACM) by CKD category using KDIGO guidelines risk classification systems. (A1) (B1): protein excretion rate (PER) cohort; (A2) (B2): protein-creatinine ratio (PCR) cohort; (A3) (B3): dipstick cohort.
942
© 2015 Asian Pacific Society of Nephrology
Validation of CKD risk categorization system
Table 5 Reclassification accuracy of the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines risk classification system for all-cause mortality, by cohort and subgroup Cohort and subgroup
PER cohort Total population Age < 60 years Age ≥ 60 years Men Women Referral at category 3† PCR cohort Total population Age < 60 years Age ≥ 60 years Men Women Referral at category 3† Dipstick cohort Total population Age < 60 years Age ≥ 60 years Men Women Referral at category 3†
Patients (Events/No Events), n
Reclassification Accuracy Events, n (% with events)
No Events, n (% without events)
Incorrect
Correct
Net
Incorrect
Correct
Net
1219 (43/1176) 882 (12/870) 337 (31/306) 634 (28/606) 585 (15/570) 1219 (43/1176)
1 (2.3) 0 (0) 1 (3.2) 1 (3.6) 0 (0) 0 (0)
14 (32.6) 5 (41.7) 9 (29.0) 7 (25.0) 7 (46.7) 10 (23.3)
13 (30.2) 5 (41.7) 8 (25.8) 6 (21.4) 7 (46.7) 10 (23.3)
456 (38.8) 365 (42.0) 91 (29.7) 259 (42.7) 197 (34.6) 189 (16.1)
47 (4.0) 15 (1.7) 32 (10.5) 30 (5.0) 17 (3.0) 0 (0)
−409 (−34.8) −350 (−40.2) −59 (−19.3) −229 (−37.8) −180 (−31.6) −189 (−16.1)
1219 (43/1176) 882 (12/870) 337 (31/306) 634 (28/606) 585 (15/570) 1219 (43/1176)
1 (2.3) 0 (0) 1 (3.2) 1 (3.6) 0 (0) 0 (0)
14 (32.6) 5 (41.7) 9 (29.0) 6 (21.4) 8 (53.3) 10 (23.3)
13 (30.2) 5 (41.7) 8 (25.8) 5 (17.9) 8 (53.3) 10 (23.3)
396 (33.7) 311 (35.8) 85 (27.8) 203 (33.5) 193 (33.9) 178 (15.1)
52 (4.4) 19 (2.2) 33 (10.8) 35 (5.8) 17 (3.0) 0 (0)
−344 (−29.3) −292 (−33.6) −52 (−17.0) −168 (−27.7) −176 (−30.9) −178 (−15.1)
1219 (43/1176) 882 (12/870) 337 (31/306) 634 (28/606) 585 (15/570) 1219 (43/1176)
2 (4.7) 0 (0) 2 (6.5) 2 (7.1) 0 (0) 0 (0)
8 (18.6) 4 (33.3) 4 (12.9) 4 (14.3) 4 (26.7) 8 (18.6)
6 (14.0) 4 (33.3) 2 (6.5) 2 (7.1) 4 (26.7) 8 (18.6)
153 (13.0) 104 (12.0) 49 (16.0) 95 (15.7) 58 (10.2) 114 (9.7)
50 (4.3) 18 (2.1) 32 (10.5) 31 (5.1) 19 (3.3) 0 (0)
−103 (−8.8) −86 (−9.9) −17 (−5.6) −64 (−10.6) −39 (−6.8) −114 (−9.7)
†Reclassification is defined as movement from referral to nonreferral (risk category 3 in the old system to risk category 1 or 2 in the new system) or vice versa.
from America and other developed countries, where persons with CKD are much more likely to die of cardiovascular disease than to experience ESRD27 and the major causes of ESRD are diabetes mellitus and hypertension. The age and baseline proteinuria levels may affect the competing outcomes of death and ESRD.28,29 Second, it is the first time that the 24 h urine PER has been applied in evaluating the predictive strength of proteinuria. The results of Cox proportional hazards model and NRI calculation in our study provide solid evidence that proteinuria measured by PER, PCR and dipstick testing are associated independently with ESRD and ACM. The findings may have some important clinical, epidemiologic and economy implications, specifically for developing countries such as China. Because measurement of PER or dipstick testing is cheaper than that of urine albumin-creatinine ratio (ACR). Third, strict training processes and vigorous quality assurance programmes were used to ensure the quality of data collection and follow-up. In 35% of the study population, baseline eGFR and proteinuria were based on multiple measurements. As such, overestimation of proteinuria and underestimation of eGFR would be rare. This study has certain limits and constraints. First, because patients without contact information or who were lost to follow-up were not included in the present study, some sort of population bias may have occurred. However, baseline characteristics of excluded patients did not differ significantly © 2015 Asian Pacific Society of Nephrology
with the study population. Second, albuminuria has been proven to be a significant independent risk factor for ESRD and ACM,30 but spot urine ACR had not been tested in this study. We could not compare albuminuria with proteinuria in respect to risk prediction of clinical adverse outcomes. Third, the specimen collection of 24 h urine is cumbersome and easy to get wrong. However, the nephrology centre is the Key department of Peking University First Hospital and there is a standard protocol of specimen collection to ensure accuracy. Moreover, it is a medium-sized, single centre CKD cohort. In conclusion, lower eGFR and higher proteinuria are risk factors for ESRD and ACM in Chinese patients. Our findings suggest that the KDIGO guidelines risk categorization system provides a more accurate risk prediction of ESRD and ACM than the KDOQI guidelines risk categorization system.
ACKNOWLEDGEMENTS This study was partially supported by the National Science Foundation of China to Li Haixia (grant no. 81101308).
REFERENCES 1. National Kidney Foundation. K/DOQI clinical practice guidelines for chronic kidney disease: Evaluation, classification, and stratification. Am. J. Kidney Dis. 2002; 39: S1-266.
943
Q Liu et al.
2. Hallan SI, Ritz E, Lydersen S, Romundstad S, Kvenild K, Orth SR. Combining GFR and albuminuria to classify CKD improves prediction of ESRD. J. Am. Soc. Nephrol. 2009; 20: 1069–77. 3. Hillege HL, Fidler V, Diercks GF et al. Urinary albumin excretion predicts cardiovascular and noncardiovascular mortality in general population. Circulation 2002; 106: 1777–82. 4. Klausen K, Borch-Johnsen K, Feldt-Rasmussen B et al. Very low levels of microalbuminuria are associated with increased risk of coronary heart disease and death independently of renal function, hypertension, and diabetes. Circulation 2004; 110: 32–5. 5. Mann JF, Gerstein HC, Pogue J, Bosch J, Yusuf S. Renal insufficiency as a predictor of cardiovascular outcomes and the impact of ramipril: The HOPE randomized trial. Ann. Intern. Med. 2001; 134: 629–36. 6. Matsushita K, Ballew SH, Astor BC et al. Cohort profile: The chronic kidney disease prognosis consortium. Int. J. Epidemiol. 2013; 42: 1660–68. 7. van der Velde M, Matsushita K, Coresh J et al. Lower estimated glomerular filtration rate and higher albuminuria are associated with all-cause and cardiovascular mortality. A collaborative meta-analysis of high-risk population cohorts. Kidney Int. 2011; 79: 1341–52. 8. Astor BC, Matsushita K, Gansevoort RT et al. Lower estimated glomerular filtration rate and higher albuminuria are associated with mortality and end-stage renal disease. A collaborative meta-analysis of kidney disease population cohorts. Kidney Int. 2011; 79: 1331–40. 9. Gansevoort RT, Matsushita K, van der Velde M et al. Lower estimated GFR and higher albuminuria are associated with adverse kidney outcomes. A collaborative meta-analysis of general and high-risk population cohorts. Kidney Int. 2011; 80: 93–104. 10. Matsushita K, van der Velde M, Astor BC et al. Association of estimated glomerular filtration rate and albuminuria with all-cause and cardiovascular mortality in general population cohorts: A collaborative meta-analysis. Lancet 2010; 375: 2073–81. 11. Wen CP, Matsushita K, Coresh J et al. Relative risks of chronic kidney disease for mortality and end-stage renal disease across races are similar. Kidney Int. 2014; 86: 819–27. 12. Hallan SI, Matsushita K, Sang Y et al. Age and association of kidney measures with mortality and end-stage renal disease. JAMA 2012; 308: 2349–60. 13. Nitsch D, Grams M, Sang Y et al. Associations of estimated glomerular filtration rate and albuminuria with mortality and renal failure by sex: A meta-analysis. BMJ 2013; 346: f324. 14. Fox CS, Matsushita K, Woodward M et al. Associations of kidney disease measures with mortality and end-stage renal disease in individuals with and without diabetes: A meta-analysis. Lancet 2012; 380: 1662–73. 15. Mahmoodi BK, Matsushita K, Woodward M et al. Associations of kidney disease measures with mortality and end-stage renal disease in individuals with and without hypertension: A meta-analysis. Lancet 2012; 380: 1649–61.
944
16. Kidney Disease: Improving Global Outcomes (KDIGO) CKD Work Group. KDIGO 2012 clinical practice guideline for the evaluation and management of chronic kidney disease. Kidney. Int. 2013; 3 (Suppl): 1–150. 17. Rodbard HW, Blonde L, Braithwaite SS et al. American Association of Clinical Endocrinologists medical guidelines for clinical practice for the management of diabetes mellitus. Endocr. Pract. 2007; 13 (Suppl 1): 1–68. 18. Mancia G, De Backer G, Dominiczak A et al. 2007 ESH-ESC practice guidelines for the management of arterial hypertension: ESH-ESC task force on the management of arterial hypertension. J. Hypertens. 2007; 25: 1751–62. 19. Kidney Disease: Improving Global Outcomes (KDIGO) CKD Work Group. KDIGO Clinical Practice Guideline for Glomerulonephritis. Kidney. Int. 2012; 2 (Suppl): 1–143. 20. Levey AS, Stevens LA, Schmid CH et al. A new equation to estimate glomerular filtration rate. Ann. Intern. Med. 2009; 150: 604–12. 21. Hemmelgarn BR, Manns BJ, Lloyd A et al. Relation between kidney function, proteinuria, and adverse outcomes. JAMA 2010; 303: 423–9. 22. Lamb EJ, MacKenzie F, Stevens PE. How should proteinuria be detected and measured? Ann. Clin. Biochem. 2009; 46: 205–17. 23. Martin DO, Austin H. Exact estimates for a rate ratio. Epidemiology 1996; 7: 29–33. 24. Pencina MJ, D’Agostino RS, D’Agostino RJ, Vasan RS. Evaluating the added predictive ability of a new marker: From area under the ROC curve to reclassification and beyond. Stat. Med. 2008; 27: 157–72, 207–12. 25. Ninomiya T, Perkovic V, de Galan BE et al. Albuminuria and kidney function independently predict cardiovascular and renal outcomes in diabetes. J. Am. Soc. Nephrol. 2009; 20: 1813–21. 26. Tonelli M, Muntner P, Lloyd A et al. Using proteinuria and estimated glomerular filtration rate to classify risk in patients with chronic kidney disease: A cohort study. Ann. Intern. Med. 2011; 154: 12–21. 27. Sarnak MJ, Levey AS, Schoolwerth AC et al. Kidney disease as a risk factor for development of cardiovascular disease: A statement from the American Heart Association Councils on Kidney in Cardiovascular Disease, High Blood Pressure Research, Clinical Cardiology, and Epidemiology and Prevention. Circulation 2003; 108: 2154–69. 28. Obi Y, Kimura T, Nagasawa Y et al. Impact of age and overt proteinuria on outcomes of stage 3 to 5 chronic kidney disease in a referred cohort. Clin. J. Am. Soc. Nephrol. 2010; 5: 1558–65. 29. O’Hare AM, Choi AI, Bertenthal D et al. Age affects outcomes in chronic kidney disease. J. Am. Soc. Nephrol. 2007; 18: 2758–65. 30. Berhane AM, Weil EJ, Knowler WC, Nelson RG, Hanson RL. Albuminuria and estimated glomerular filtration rate as predictors of diabetic end-stage renal disease and death. Clin. J. Am. Soc. Nephrol. 2011; 6: 2444–51.
© 2015 Asian Pacific Society of Nephrology
Nephrology 20 (2015) 936–944
Original Article
Validation of chronic kidney disease risk categorization system in Chinese patients with kidney disease: A cohort study QINGYAN LIU,1 JICHENG LV,2,3,4 HAIXIA LI,1 LILI JIAO,1 HONGYUN YANG,1 YINAN SONG1 and GUOBIN XU1 1 Department of Clinical Laboratory, 2Division of Nephrology and Institute of Nephrology, Peking University First Hospital, 3Key Laboratory of Renal Disease, Ministry of Health of China, and 4Key Laboratory of Chronic Kidney Disease Prevention and Treatment (Peking University), Ministry of Education, Beijing, China
KEY WORDS: chronic kidney disease, glomerular filtration rate, prognosis, proteinuria. Correspondence Prof Guobin Xu, No. 8, Xishiku Street, Xicheng District, Beijing 100034, China. Email: [email protected] Accepted for publication 27 May 2015. Accepted manuscript online 1 June 2015. doi:10.1111/nep.12528 Conflict of interest: The authors declare no conflict of interest.
SUMMARY AT A GLANCE This paper provides a validation of the KDIGO risk stratification in a Chinese population living in China, based on a combination of eGFR and proteinuria.
ABSTRACT Aim: To validate the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines risk stratification system based on the combination of estimated glomerular filtration rate (eGFR) and proteinuria. Methods: This was a cohort study. A total of 1219 study population were recruited. Estimated GFR and proteinuria measured by using 24 h urine protein excretion rate (PER) were predictors. Adverse outcomes included all-cause mortality (ACM) and end-stage renal disease (ESRD). Follow-up was done by regular visit, telephone interview and electronic medical records. Results: Over a median follow-up of 4.6 years, 153 (12.6%) and 43 (3.5%) patients experienced ESRD and ACM, respectively. On multivariable analysis, the adjusted hazard ratio for ESRD and ACM (compared with patients with eGFR > 60 mL/min per 1.73 m2) was of 29.8 and 3.6 for those with eGFR of 15–29 mL/min per 1.73 m2, respectively. The adjusted hazard ratio for ESRD and ACM (compared with patients with PER < 150 mg/24h) was of 15.9 and 3.9 for those with PER > 500 mg/24h. Higher KDIGO guidelines risk categories (indicating lower eGFR or higher proteinuria) were associated with a graded increase in the risk for the ESRD (P < 0.001) and ACM (P < 0.001). Reclassification of KDIGO guidelines risk categories yielded net reclassification improvements for those with ESRD or ACM event (NRIevents) of 33.3% or 30.2%. Conclusion: Lower eGFR and higher proteinuria are risk factors for ESRD and ACM in Chinese patients. The KDIGO guidelines risk categorization system assigned patients who went on to have the event to more appropriate CKD risk categories.
The Kidney Disease Outcomes Quality Initiative (KDOQI) guidelines1 risk categorization system, which relies almost exclusively on estimated glomerular filtration rate (eGFR), does not incorporate information about how the presence and severity of proteinuria might affect prognosis for important clinical outcomes in each chronic kidney disease (CKD) risk category.2–5 The vast body of evidence provided by the CKD Prognosis Consortium6 demonstrated that both eGFR and albuminuria were associated with adverse kidney outcomes, and all-cause and cardiovascular mortality, independent of traditional cardiovascular risk factors and independent of each other and despite inclusion of diverse study 936
populations.7–15 The Kidney Disease: Improving Global Outcomes (KDIGO) guidelines,16 therefore recommended CKD staging to be based on both eGFR and proteinuria and proposed four risk categories according to the combination of eGFR categories with proteinuria categories. However, data for validation of the KDIGO guidelines risk categorization system in Chinese patients is limited and the prognostic implication of stratifying patients by combining eGFR categories with 24 h urine protein excretion rate (PER) categories has not yet been investigated among CKD patients with a broad spectrum of underlying aetiologies. We validated that the KDIGO guidelines risk categorization system more © 2015 Asian Pacific Society of Nephrology
Validation of CKD risk categorization system
accurately predicts risk for all-cause mortality (ACM) and end-stage renal disease (ESRD) than the KDOQI guidelines risk categorization system in this study.
METHODS Study population A total of 3043 patients with laboratory tests of 24 h endogenous creatinine clearance rate (Ccr) and 24 h urine PER on the same day between May 2009 with May 2010 in a tertiary nephrology centre were continuously enrolled. After we excluded 155 patients with previous ESRD or an initial eGFR less than 15 mL/min per 1.73 m2 (on the assumption that they would reach the end of follow-up), 251 patients who were younger than 18 years and 214 pregnant patients, and 1058 patients without contact information, a total of 1365 patients entered the follow-up. Because of unknown follow-up status, 146 patients were missed. Data were available for 1219 patients with kidney disease. Flow diagram of the study population is shown in Figure 1. Data on demographics, history of diabetes mellitus, hypertension, co-morbidities and current drug intake were recorded by a standardized questionnaire. Laboratory values and contact information were determined from the administrative data files in the laboratory information system and hospital information system of Peking University First Hospital. Diabetes mellitus, hypertension, or other comorbid conditions was identified by American Association of Clinical Endocrinologists medical guidelines for clinical practice for the management of diabetes mellitus,17 2007 ESH-ESC Practice Guidelines for the Management of Arterial Hypertension,18 or KDIGO Clinical Practice Guidelines for Glomerulonephritis.19 A specialist in nephrology confirmed the aetiology of CKD by medical history, clinical manifestation, laboratory examination, and kidney biopsy. This study was approved by the Medical Ethics Committee of Peking University First Hospital. All patients provided oral or written consent to their participation in the follow-up.
Assessment of kidney function and proteinuria Estimated GFR for each patient was calculated by using the Chronic Kidney Disease-Epidemiology Collaboration (CKD-EPI) two-level race creatinine equation.20 Serum creatinine was measured in blood samples after an overnight fast, using the Jaffé method (CV ≤ 5%), which can be traceable to an isotope dilution mass spectrometry reference. For patients with multiple creatinine measurements, the mean of all measures taken within 4 months of the first measurement in the follow-up period was used (on the assumption that abnormalities of kidney function would present for >3 months), with the date of the last measurement in that 4-month period used as the baseline date. Proteinuria was measured on the basis of 24 h urine measurements by using PER, PCR and urine dipstick testing, respectively. PER was the urinary total protein concentration multiplied by volumes of 24 h urine. Urinary total protein concentration was measured by pyrogallol red colorimetric method. PCR was the urinary total protein concentration divided by urine creatinine concentration in 24 h urine. Urine creatinine concentration was determined using the Jaffé method, and then converted the concentration unit of mmol/L to g/L to calculate PCR. Urinary total protein concentration (≤100 mg/L, CV < 20%; >100 mg/L, CV < 7%) and urinary creatinine concentration (CV ≤ 5%) were measured on Hitachi 7180 Biochemical analysis system. Urine dipstick testing was measured by bromophenol blue assay. PER, PCR and urine dipstick measurements tested on the same day as Ccr measurement were used to establish baseline proteinuria. For patients with multiple PER, PCR and urine dipstick measurements, the median (using ordinal numbers for dipstick protein categories)21 of all measures taken within 4 months of the first measurement in the follow-up period was used (on the assumption that kidney damage would present for >3 months).
CKD risk category Baseline eGFR for each patient was categorized as 60 or greater, 45 to 59, 30 to 44, and 15 to 29 mL/min per 1.73 m2. Proteinuria among patients was classified as normal (PER < 150 mg/24h, PCR < 150 mg/g or urine dipstick negative), mild (PER 150 to 500 mg/24h, PCR 150 to 500 mg/g or urine dipstick trace or 1+), or heavy (PER > 500 mg/24h, PCR > 500 mg/g or urine dipstick ≥2+).16,22 Each patient was placed into a proteinuria category based on PER, PCR and dipstick results, respectively. Similarly, each patient was allocated a CKD risk category for each combination of eGFR category and proteinuria categories allocated by PER, PCR and dipstick results, respectively. CKD risk categories (0, 1, 2 and 3) were in accordance with KDOQI guidelines1 and 2012 KDIGO guidelines.16 Risk category 0, 1, 2 and 3 reflect the low, medium, high and very high risk of progression, respectively.
Adverse outcome assessment
Fig. 1 Flow diagram of the study population. © 2015 Asian Pacific Society of Nephrology
All patients were invited to regularly visit the outpatient department, and information on laboratory values was obtained from electronic medical records. Patients who did not attend regular visits were contacted via telephone. Trained research personnel making the telephone calls were blinded to patients’ baseline clinical status. Patients were asked standardized follow-up questions at the end of 937
Q Liu et al.
follow-up. Outcomes data collected included ACM and ESRD. ACM was defined as death from any cause that was confirmed by either a death certificate from the local citizen registry or the electronic medical record of Peking University First Hospital. When no official documentation was available, case fatality was decided if death was reported by different proxies. ESRD was defined as registration for long-term dialysis or renal transplantation. Cases of ESRD were cross-checked with the treating hospitals to ensure the accuracy of diagnosis. Date of ESRD initiation or death was determined from telephone interview and electronic medical records. Patients who experienced ESRD and then died were considered to have experienced both adverse outcomes, with the date for each event used to calculate follow-up time. All outcomes were assessed from the baseline date until study end (31 August 2014).
Statistical analysis We validated the KDIGO guidelines risk categorization system by using PER, PCR and dipstick cohort. PER and PCR were converted to a log scale, and then compared with a scatterplot, using both Lowess (locally weighted scatterplot smoothing) and Deming methods to fit the regression. Spearman correlation coefficients were calculated to assess the correlation between PER and PCR in the study population. PER, PCR or urine dipstick result was used to cross-tabulating together with baseline eGFR, respectively. Rates of adverse outcomes (ESRD and ACM) were calculated by Poisson regression for each of 12 groups defined by the four eGFR and three proteinuria categories, with output expressed per 1000 person-years. The association between eGFR or proteinuria and adverse outcomes was analyzed by the Cox proportional hazard model. Four CKD categories ranging from 0 (no CKD) to 3 (most severe CKD) were formed by KDIGO guidelines risk categorization system. Incidence rates of ESRD and ACM in four CKD categories were calculated as cumulative incidence and compared using the rate ratio.23 The reclassification accuracy of participants with and without adverse outcomes using KDIGO versus KDOQI guidelines risk categorization system was quantified using the net reclassification improvement (NRI),24 which was defined as the proportions of ‘clinically correct’ reclassification minus the proportions of ‘clinically incorrect’ reclassification. Specifically, the ‘clinically correct’ reclassification was the reclassification of people with adverse outcomes to higher versus lower categories and for those without adverse outcomes to lower versus higher categories. The consistency of these findings was also assessed in subgroups stratified by age ( 60 mL/min per 1.73 m2) was of 29.8 (95% CI: 16.1–55.1) and 3.6 (95% CI: 1.3–10.1) for those with eGFR of 15–29 mL/min per 1.73 m2, respectively (Table 2). Likewise, higher proteinuria was associated with higher rate of ESRD and ACM overall and within every estimated GFR category. The adjusted hazard ratio for ESRD and ACM (compared with patients with PER < 150 mg/24 h) was of 15.9 (95% CI: 2.2–116.8) and 3.9 (95% CI: 1.1–14.7) for those with PER > 500 mg/24 h, respectively (Table 3). Results were similar in the analysis of PCR (Fig. 3A2,B2) and dipstick cohort (Fig. 3A3,B31).
Cumulative incidences of ESRD and ACM by CKD category In the PER cohort, the cumulative incidences of ESRD in CKD category 0–3 were 0, 1.6, 10.5 and 116.4 per 1000 person-years, respectively and the cumulative incidences of ACM in CKD category 0–3 were 3.4, 4.0, 4.5 and 23.8 per 1000 person-years, respectively. In the PCR cohort, the cumulative incidences of ESRD in CKD category 0–3 were 0, 1.5, 12.4 and 121.5 per 1000 person-years, respectively, and the cumulative incidences of ACM in CKD category 0–3 were 3.0, 3.8, 5.3 and 24.8 per 1000 person-years, respectively. In the dipstick cohort, the cumulative incidences of ESRD in CKD category 0–3 were 0, 4.7, 28.5, 144.2 per 1000 personyears, respectively and the cumulative incidences of ACM in CKD category 0–3 were 3.2, 4.7, 4.3 and 30.5 per 1000 © 2015 Asian Pacific Society of Nephrology
Details for reclassification accuracy of the KDIGO risk classification system for ESRD are listed in Table 4. In the PER cohort, Of 153 patients who suffered ESRD events during the follow-up, one (0.7%) patient was incorrectly reclassified to a lower risk category and 53 (34%) patients were correctly reclassified to a higher risk category using the KDIGO guidelines risk classification system, yielding an NRIevents of 33.3%. In contrast, of 1066 patients who did not experience ESRD events, 47 (4.4%) patients were correctly reclassified to a lower risk category and 418 (39.2%) patients were incorrectly reclassified to a higher risk category, resulting in a NRInon-events of −34.8%. Similarly, the NRIevents and NRInon-events in the PCR and dipstick cohorts were 33.3% and −28.7%, 19.0% and −7.5%, respectively. Details for reclassification accuracy of the KDIGO guidelines risk classification system for ACM are listed in Table 5. In the PER cohort, of 43 patients who suffered ACM events during the follow-up, one (2.3%) patient was incorrectly reclassified to a lower risk category and 14 (32.6%) patients were correctly reclassified to a higher risk category using the KDIGO risk classification system, yielding an NRIevents of 30.2%. In contrast, of 1176 patients who did not experience ACM events, 47 (4.0%) patients were correctly reclassified to a lower risk category and 456 (38.8%) patients were incorrectly reclassified to a higher risk category, resulting in a NRInon-events of −34.8%. Similarly, the NRIevents and NRInon-events in PCR and dipstick cohort were 30.2% and −29.3%, 14.0% and −8.8%, respectively.
DISCUSSION This cohort study is the first to validate the KDIGO guidelines risk categorization system in Chinese CKD patients with a broad spectrum of underlying aetiologies and prove the performance of 24 h urine PER, PCR and dipstick urinalysis in predicting risk of ESRD and ACM. Overall, our results showed that decreased eGFR as calculated by CKD-EPI equation and more proteinuria as measured by 24 h urine PER, PCR, and dipstick urinalysis were associated independently with increased risk of ESRD and ACM (Fig. 3, Tables 2,3). However, the predictive strength of PCR and PER appeared to be stronger than that of the dipstick urinalysis (Tables 4,5). Moreover, the present study also showed that KDIGO guidelines risk categorization system appeared to be more appropriate than KDOQI guidelines risk categorization system in Chinese patients with kidney disease according to their risk of ESRD and ACM. KDIGO guidelines risk 939
Q Liu et al.
Fig. 3 Unadjusted incidence rates (per 1000 person-years) for (A) end-stage renal disease (ESRD) (B) all-cause mortality (ACM) in the protein excretion rate (PER), PCR and dipstick cohort. (A1) (B1): PER cohort; (A2) (B2): protein-creatinine ratio (PCR) cohort; (A3) (B3): dipstick cohort.
Table 2 Adjusted† hazard ratio (95% confidence interval) for end-stage renal disease and all-cause mortality, by proteinuria glomerular filtration rate (GFR0 category eGFR (mL/min per 1.73 m2)
ESRD ACM
≥60
45–59
30–44
15–29
Ref. Ref.
3.6 (1.7, 7.5) P = 0.001 0.8 (0.2, 2.5) P = 0.671
6.6 (3.2, 13.6) P < 0.001 0.7 (0.2, 2.9) P = 0.652
29.8 (16.1, 55.1) P < 0.001 3.6 (1.3, 10.1) P = 0.016
†Adjusted for age, sex, body mass index (BMI), diabetes mellitus, hypertension and proteinuria. ACM, all-cause mortality; ESRD, end-stage renal disease.
940
© 2015 Asian Pacific Society of Nephrology
Validation of CKD risk categorization system
Table 3 Adjusted† hazard ratio (95% confidence interval) for end-stage renal disease (ESRD) and all-cause mortality (ACM), by proteinuria category Proteinuria
PER cohort ESRD ACM PCR cohort ESRD ACM Dipstick cohort ESRD ACM
Normal
Mild
Heavy
Ref. Ref.
5.1 (0.6, 40.5) P = 0.124 1.6 (0.3, 7.2) P = 0.562
15.9 (2.2, 116.8) P = 0.007 3.9 (1.1, 14.7) P = 0.042
Ref. Ref.
7.0 (0.9, 55.0) P = 0.064 3.2 (0.8, 13.1) P = 0.110
19.6 (2.7, 144.1) P = 0.003 3.6 (0.9, 13.5) P = 0.043
Ref. Ref.
5.7 (1.4, 23.9) P = 0.016 1.8 (0.6, 5.9) P = 0.310
13.7 (3.1, 60.1) P = 0.001 4.3 (1.0, 19.3) P = 0.035
†Adjusted for age, sex, body mass index (BMI), diabetes mellitus, hypertension and estimated glomerular filtration rate (eGFR).
Table 4 Reclassification accuracy of the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines risk classification system for end-stage renal disease, by cohort and subgroup Cohort and subgroup
PER cohort Total population Age < 60 years Age ≥ 60 years Men Women Referral at category 3† PCR cohort Total population Age < 60 years Age ≥ 60 years Men Women Referral at category 3† Dipstick cohort Total population Age < 60 years Age ≥ 60 years Men Women Referral at category 3†
Patients (Events/No Events), n
Reclassification Accuracy Events, n (% with events)
No Events, n (% without events)
Incorrect
Correct
Net
Incorrect
Correct
Net
1219 (153/1066) 882 (97/785) 337 (56/281) 634 (94/540) 585 (59/526) 1219 (153/1066)
1 (0.7) 0 (0) 1 (1.8) 0 (0) 1 (1.7) 0 (0)
52 (34.0) 40 (41.2) 12 (21.4) 34 (36.2) 18 (30.5) 40 (26.1)
51 (33.3) 40 (41.2) 11 (19.6) 34 (36.2) 17 (28.8) 40 (26.1)
418 (39.2) 330 (42.0) 88 (31.3) 232 (43.0) 186 (35.4) 159 (14.9)
47 (4.4) 15 (1.9) 32 (11.4) 31 (5.7) 16 (3.0) 0 (0)
−371 (−34.8) −315 (−40.1) −56 (−19.9) −201 (−37.2) −170 (−32.3) −159 (−14.9)
1219 (153/1066) 882 (97/785) 337 (56/281) 634 (94/540) 585 (59/526) 1219 (153/1066)
1 (0.7) 0 (0) 1 (1.8) 0 (0) 1 (1.7) 0 (0)
52 (34.0) 40 (41.2) 12 (21.4) 34 (36.2) 18 (30.5) 40 (26.1)
51 (33.3) 40 (41.2) 11 (19.6) 34 (36.2) 17 (28.8) 40 (26.1)
358 (33.6) 276 (35.2) 82 (29.2) 175 (32.4) 183 (34.8) 148 (13.9)
52 (4.9) 19 (2.4) 33 (11.7) 36 (6.7) 16 (3.0) 0 (0)
−306 (−28.7) −257 (−32.7) −49 (−17.4) −139 (−25.7) −167 (−31.8) −148 (−13.9)
1219 (153/1066) 882 (97/785) 337 (56/281) 634 (94/540) 585 (59/526) 1219 (153/1066)
2 (1.3) 0 (0) 2 (3.6) 1 (1.1) 1 (1.7) 0 (0)
31 (20.3) 25 (25.8) 6 (10.7) 22 (23.4) 9 (15.3) 26 (17.0)
29 (19.0) 25 (25.8) 4 (7.1) 21 (22.3) 8 (13.6) 26 (17.0)
130 (12.2) 83 (10.6) 47 (16.7) 77 (14.3) 53 (10.1) 96 (9.0)
50 (4.7) 18 (2.3) 32 (11.4) 32 (5.9) 18 (3.4) 0 (0)
−80 (−7.5) −65 (−8.3) −15 (−5.3) −45 (−8.3) −35 (−6.7) −96 (−9.0)
†Reclassification is defined as movement from referral to nonreferral (risk category 3 in the old system to risk category 1 or 2 in the new system) or vice versa.
categorization system reclassified more of the patients who went on to have an outcome event to more advanced CKD risk categories, and correct reclassification was more likely for the ESRD than for ACM (Tables 4,5). The KDIGO guidelines risk categorization system has important clinical implications for identifying additional patients who are in need of closer follow-up and more aggressive treatment. The KDIGO guidelines risk categorization system was developed predominantly on the basis of a white population.6 However, whether the KDIGO guidelines risk categorization system more accurately predicts risks for ACM and © 2015 Asian Pacific Society of Nephrology
ESRD than the KDOQI guidelines risk categorization system in Chinese patients with kidney disease has not been tested before. Our study complements and extends the previous studies25,26 in Chinese patients with kidney disease, and supports recent recommendations defining chronic kidney disease and stratifying subsequent risks based on both decreased GFR and proteinuria. This study has several advantages. First, our analysis was performed in Chinese patients with CKD, which was predominantly secondary to glomerular disease (45%). The overall incident of ESRD is higher than ACM. It is different 941
Q Liu et al.
Fig. 4 Cumulative incidences of (A) end-stage renal disease (ESRD)and (B) all-cause mortality (ACM) by CKD category using KDIGO guidelines risk classification systems. (A1) (B1): protein excretion rate (PER) cohort; (A2) (B2): protein-creatinine ratio (PCR) cohort; (A3) (B3): dipstick cohort.
942
© 2015 Asian Pacific Society of Nephrology
Validation of CKD risk categorization system
Table 5 Reclassification accuracy of the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines risk classification system for all-cause mortality, by cohort and subgroup Cohort and subgroup
PER cohort Total population Age < 60 years Age ≥ 60 years Men Women Referral at category 3† PCR cohort Total population Age < 60 years Age ≥ 60 years Men Women Referral at category 3† Dipstick cohort Total population Age < 60 years Age ≥ 60 years Men Women Referral at category 3†
Patients (Events/No Events), n
Reclassification Accuracy Events, n (% with events)
No Events, n (% without events)
Incorrect
Correct
Net
Incorrect
Correct
Net
1219 (43/1176) 882 (12/870) 337 (31/306) 634 (28/606) 585 (15/570) 1219 (43/1176)
1 (2.3) 0 (0) 1 (3.2) 1 (3.6) 0 (0) 0 (0)
14 (32.6) 5 (41.7) 9 (29.0) 7 (25.0) 7 (46.7) 10 (23.3)
13 (30.2) 5 (41.7) 8 (25.8) 6 (21.4) 7 (46.7) 10 (23.3)
456 (38.8) 365 (42.0) 91 (29.7) 259 (42.7) 197 (34.6) 189 (16.1)
47 (4.0) 15 (1.7) 32 (10.5) 30 (5.0) 17 (3.0) 0 (0)
−409 (−34.8) −350 (−40.2) −59 (−19.3) −229 (−37.8) −180 (−31.6) −189 (−16.1)
1219 (43/1176) 882 (12/870) 337 (31/306) 634 (28/606) 585 (15/570) 1219 (43/1176)
1 (2.3) 0 (0) 1 (3.2) 1 (3.6) 0 (0) 0 (0)
14 (32.6) 5 (41.7) 9 (29.0) 6 (21.4) 8 (53.3) 10 (23.3)
13 (30.2) 5 (41.7) 8 (25.8) 5 (17.9) 8 (53.3) 10 (23.3)
396 (33.7) 311 (35.8) 85 (27.8) 203 (33.5) 193 (33.9) 178 (15.1)
52 (4.4) 19 (2.2) 33 (10.8) 35 (5.8) 17 (3.0) 0 (0)
−344 (−29.3) −292 (−33.6) −52 (−17.0) −168 (−27.7) −176 (−30.9) −178 (−15.1)
1219 (43/1176) 882 (12/870) 337 (31/306) 634 (28/606) 585 (15/570) 1219 (43/1176)
2 (4.7) 0 (0) 2 (6.5) 2 (7.1) 0 (0) 0 (0)
8 (18.6) 4 (33.3) 4 (12.9) 4 (14.3) 4 (26.7) 8 (18.6)
6 (14.0) 4 (33.3) 2 (6.5) 2 (7.1) 4 (26.7) 8 (18.6)
153 (13.0) 104 (12.0) 49 (16.0) 95 (15.7) 58 (10.2) 114 (9.7)
50 (4.3) 18 (2.1) 32 (10.5) 31 (5.1) 19 (3.3) 0 (0)
−103 (−8.8) −86 (−9.9) −17 (−5.6) −64 (−10.6) −39 (−6.8) −114 (−9.7)
†Reclassification is defined as movement from referral to nonreferral (risk category 3 in the old system to risk category 1 or 2 in the new system) or vice versa.
from America and other developed countries, where persons with CKD are much more likely to die of cardiovascular disease than to experience ESRD27 and the major causes of ESRD are diabetes mellitus and hypertension. The age and baseline proteinuria levels may affect the competing outcomes of death and ESRD.28,29 Second, it is the first time that the 24 h urine PER has been applied in evaluating the predictive strength of proteinuria. The results of Cox proportional hazards model and NRI calculation in our study provide solid evidence that proteinuria measured by PER, PCR and dipstick testing are associated independently with ESRD and ACM. The findings may have some important clinical, epidemiologic and economy implications, specifically for developing countries such as China. Because measurement of PER or dipstick testing is cheaper than that of urine albumin-creatinine ratio (ACR). Third, strict training processes and vigorous quality assurance programmes were used to ensure the quality of data collection and follow-up. In 35% of the study population, baseline eGFR and proteinuria were based on multiple measurements. As such, overestimation of proteinuria and underestimation of eGFR would be rare. This study has certain limits and constraints. First, because patients without contact information or who were lost to follow-up were not included in the present study, some sort of population bias may have occurred. However, baseline characteristics of excluded patients did not differ significantly © 2015 Asian Pacific Society of Nephrology
with the study population. Second, albuminuria has been proven to be a significant independent risk factor for ESRD and ACM,30 but spot urine ACR had not been tested in this study. We could not compare albuminuria with proteinuria in respect to risk prediction of clinical adverse outcomes. Third, the specimen collection of 24 h urine is cumbersome and easy to get wrong. However, the nephrology centre is the Key department of Peking University First Hospital and there is a standard protocol of specimen collection to ensure accuracy. Moreover, it is a medium-sized, single centre CKD cohort. In conclusion, lower eGFR and higher proteinuria are risk factors for ESRD and ACM in Chinese patients. Our findings suggest that the KDIGO guidelines risk categorization system provides a more accurate risk prediction of ESRD and ACM than the KDOQI guidelines risk categorization system.
ACKNOWLEDGEMENTS This study was partially supported by the National Science Foundation of China to Li Haixia (grant no. 81101308).
REFERENCES 1. National Kidney Foundation. K/DOQI clinical practice guidelines for chronic kidney disease: Evaluation, classification, and stratification. Am. J. Kidney Dis. 2002; 39: S1-266.
943
Q Liu et al.
2. Hallan SI, Ritz E, Lydersen S, Romundstad S, Kvenild K, Orth SR. Combining GFR and albuminuria to classify CKD improves prediction of ESRD. J. Am. Soc. Nephrol. 2009; 20: 1069–77. 3. Hillege HL, Fidler V, Diercks GF et al. Urinary albumin excretion predicts cardiovascular and noncardiovascular mortality in general population. Circulation 2002; 106: 1777–82. 4. Klausen K, Borch-Johnsen K, Feldt-Rasmussen B et al. Very low levels of microalbuminuria are associated with increased risk of coronary heart disease and death independently of renal function, hypertension, and diabetes. Circulation 2004; 110: 32–5. 5. Mann JF, Gerstein HC, Pogue J, Bosch J, Yusuf S. Renal insufficiency as a predictor of cardiovascular outcomes and the impact of ramipril: The HOPE randomized trial. Ann. Intern. Med. 2001; 134: 629–36. 6. Matsushita K, Ballew SH, Astor BC et al. Cohort profile: The chronic kidney disease prognosis consortium. Int. J. Epidemiol. 2013; 42: 1660–68. 7. van der Velde M, Matsushita K, Coresh J et al. Lower estimated glomerular filtration rate and higher albuminuria are associated with all-cause and cardiovascular mortality. A collaborative meta-analysis of high-risk population cohorts. Kidney Int. 2011; 79: 1341–52. 8. Astor BC, Matsushita K, Gansevoort RT et al. Lower estimated glomerular filtration rate and higher albuminuria are associated with mortality and end-stage renal disease. A collaborative meta-analysis of kidney disease population cohorts. Kidney Int. 2011; 79: 1331–40. 9. Gansevoort RT, Matsushita K, van der Velde M et al. Lower estimated GFR and higher albuminuria are associated with adverse kidney outcomes. A collaborative meta-analysis of general and high-risk population cohorts. Kidney Int. 2011; 80: 93–104. 10. Matsushita K, van der Velde M, Astor BC et al. Association of estimated glomerular filtration rate and albuminuria with all-cause and cardiovascular mortality in general population cohorts: A collaborative meta-analysis. Lancet 2010; 375: 2073–81. 11. Wen CP, Matsushita K, Coresh J et al. Relative risks of chronic kidney disease for mortality and end-stage renal disease across races are similar. Kidney Int. 2014; 86: 819–27. 12. Hallan SI, Matsushita K, Sang Y et al. Age and association of kidney measures with mortality and end-stage renal disease. JAMA 2012; 308: 2349–60. 13. Nitsch D, Grams M, Sang Y et al. Associations of estimated glomerular filtration rate and albuminuria with mortality and renal failure by sex: A meta-analysis. BMJ 2013; 346: f324. 14. Fox CS, Matsushita K, Woodward M et al. Associations of kidney disease measures with mortality and end-stage renal disease in individuals with and without diabetes: A meta-analysis. Lancet 2012; 380: 1662–73. 15. Mahmoodi BK, Matsushita K, Woodward M et al. Associations of kidney disease measures with mortality and end-stage renal disease in individuals with and without hypertension: A meta-analysis. Lancet 2012; 380: 1649–61.
944
16. Kidney Disease: Improving Global Outcomes (KDIGO) CKD Work Group. KDIGO 2012 clinical practice guideline for the evaluation and management of chronic kidney disease. Kidney. Int. 2013; 3 (Suppl): 1–150. 17. Rodbard HW, Blonde L, Braithwaite SS et al. American Association of Clinical Endocrinologists medical guidelines for clinical practice for the management of diabetes mellitus. Endocr. Pract. 2007; 13 (Suppl 1): 1–68. 18. Mancia G, De Backer G, Dominiczak A et al. 2007 ESH-ESC practice guidelines for the management of arterial hypertension: ESH-ESC task force on the management of arterial hypertension. J. Hypertens. 2007; 25: 1751–62. 19. Kidney Disease: Improving Global Outcomes (KDIGO) CKD Work Group. KDIGO Clinical Practice Guideline for Glomerulonephritis. Kidney. Int. 2012; 2 (Suppl): 1–143. 20. Levey AS, Stevens LA, Schmid CH et al. A new equation to estimate glomerular filtration rate. Ann. Intern. Med. 2009; 150: 604–12. 21. Hemmelgarn BR, Manns BJ, Lloyd A et al. Relation between kidney function, proteinuria, and adverse outcomes. JAMA 2010; 303: 423–9. 22. Lamb EJ, MacKenzie F, Stevens PE. How should proteinuria be detected and measured? Ann. Clin. Biochem. 2009; 46: 205–17. 23. Martin DO, Austin H. Exact estimates for a rate ratio. Epidemiology 1996; 7: 29–33. 24. Pencina MJ, D’Agostino RS, D’Agostino RJ, Vasan RS. Evaluating the added predictive ability of a new marker: From area under the ROC curve to reclassification and beyond. Stat. Med. 2008; 27: 157–72, 207–12. 25. Ninomiya T, Perkovic V, de Galan BE et al. Albuminuria and kidney function independently predict cardiovascular and renal outcomes in diabetes. J. Am. Soc. Nephrol. 2009; 20: 1813–21. 26. Tonelli M, Muntner P, Lloyd A et al. Using proteinuria and estimated glomerular filtration rate to classify risk in patients with chronic kidney disease: A cohort study. Ann. Intern. Med. 2011; 154: 12–21. 27. Sarnak MJ, Levey AS, Schoolwerth AC et al. Kidney disease as a risk factor for development of cardiovascular disease: A statement from the American Heart Association Councils on Kidney in Cardiovascular Disease, High Blood Pressure Research, Clinical Cardiology, and Epidemiology and Prevention. Circulation 2003; 108: 2154–69. 28. Obi Y, Kimura T, Nagasawa Y et al. Impact of age and overt proteinuria on outcomes of stage 3 to 5 chronic kidney disease in a referred cohort. Clin. J. Am. Soc. Nephrol. 2010; 5: 1558–65. 29. O’Hare AM, Choi AI, Bertenthal D et al. Age affects outcomes in chronic kidney disease. J. Am. Soc. Nephrol. 2007; 18: 2758–65. 30. Berhane AM, Weil EJ, Knowler WC, Nelson RG, Hanson RL. Albuminuria and estimated glomerular filtration rate as predictors of diabetic end-stage renal disease and death. Clin. J. Am. Soc. Nephrol. 2011; 6: 2444–51.
© 2015 Asian Pacific Society of Nephrology
