Development Of Minimal Clinically Important Difference For Proteinuria In Adult With Diabetic Kidney Disease

ObjectiveThis study attempted to answer the question that what is the minimal difference value of proteinuria which could be considered as clinically significant in patients with diabetic kidney disease(DKD)who are treated with Chinese herbal medicine(CHM).Therefore,the main purpose of this study was to develop the minimal clinically important difference(MCID)of proteinuria in DKD patients after receiving CHM treatment based on a real-world cohort.The MCID of proteinuia would provide a reasonable cut-off for evaluating the clinical significance of the results in trials of assessing the effects of CHM for DKD,and also provide a reference for determination of the margin in sample size calculation.MethodsThis study had been approved by the Ethics Committee of Guangdong Provincial Hospital of Chinese Medicine(approval number:BE2020-129-01).The study was composed of two parts.The first part was a systematic review of double blinded,randomized and placebo controlled trials which assessed effects of CHM for DKD.Meta analysis was performed to investigate changes of proteinuria in terms of 24-hour urinary protein excretion(24h UPE),24-hour urinary albumin excretion(24h UAE),urine protein/creatinine ratio(PCR)and urinary albumin/creatinine ratio(ACR).The second part was the calculation of MDC and MCID of proteinuria by using data from a real-world cohort.The cohort that consisted of 145 DKD patients in CKD stage 1 and 2who were treated with CHM and followed up in Guangdong Provincial Hospital of Chinese Medicine.MDCs of proteinuria were calculated via distribution-based methods by calculating effect size(ES),standardized response mean(SRM),standard deviation(SD)and standard error of measurement(SEM).Pooled MDC were the mean of the four results.MCIDs of proteinuria were calculated by anchor-based methods.End stage renal disease(ESRD),doubling of serum creatinine and a 40%reduction in glomerular filtration rate(GFR)(40%GFR reduction)were the anchor candidates.Only when the change of proteinuria and the candidates had an accepteble correlation(i.e.correlation coefficient more than 0.3),the candidate was selected as anchor.The receiver operator characteristic(ROC)curve and the mean difference method were used.When the area under the curve(AUC)was greater than 0.7,the results were used as preliminary MCIDs.Pooled MCIDs were the mean of the results.The anchor-based methods were used to determine the final MCID.The distribution-based methods and meta-analysis results were considered as supportive results for diterming the final MCID.When preliminary MCIDs were greater than MDC,preliminary MCIDs were used as the final MCID.When preliminary MCIDs were less than MDCs,MDCs and the results of Meta analysis of RCT were considered to determine the final MCID.Results Part oneA comprehensive search yielded more than 50,000 Chinese and English literatures.Eventually,18 double-blinded,randomized,placebo-controlled trials of CHM for DKD were included.A total of 2378 DKD patients with an average age of 55.1 years old.The baseline 24h UPE and 24h UAE were 961mg/24h and 186.88mg/24h respectively.Meta analysis results showed that when CHM added to conventional therapies which did not include renin-aldosterone system blockers,24UPE decreased by 285.40mg/24h(P<0.05)and 24h UAE decreased by 83.08mg/24h(P<0.05)with treatments of 8-12 weeks.When CHM added to conventional therapies which included renin-aldosterone system blockers,24h UPE decreased by 111.89mg/24h(P<0.05)and 24h UAE decreased by 102.89mg/24h(P<0.05).Part twoTotally,145 DKD patients in CKD stage 1 and 2 presenting proteinuria or albuminuria were included in the cohort.The mean follow up was 3.35 years.The median age was 57 years old.The baseline serum creatinine and e GFR were 69?mol/L and 85ml/min/1.73m~2,24h UPE and 24h UAE were 933mg/24h and544mg/24h.As inadequate report of 24h UAE,ACR and PCR,MCID estimation was only applied to 24h UPE.Results of Pearson correlation between changes of 24h UPE and serum creatinine or e GFR were statistically significant,and the correlation coefficient of 2-3 year changes of 24h UPE and serum creatinine and e GFR were greater than 0.3.Therefore,3 candidates were all selected as anchors.MCIDs estimated by the mean difference methodThere was no statistical significance between the progressive group and non-progressive group categorized by each anchor for 3-6 months and 6 months to 1 year 24h UPE.The difference of 1-2 years 24h UPE change between groups categorized by 40%e GFR reduction was statistically significant.Estimated MCIDs were-1045mg/24h for the entire sample,-1552mg/24h for subgroup of CKD stage 1,and-953mg/24h for subgroup of CKD stage 2.The difference of2-3 years 24h UPE change between groups categorized by each anchor were statistical significant.The pooled estimated MCIDs were-2474mg/24h for the entire sample,-1838mg/24h for subgroup of CKD stage 2,and-1686mg/24h for subgroup of non-nephrotic proteinuria.MCID estimated by ROCROC was only applied to calculate the MCID of 2-3 years 24h UPE,because only the change of 24h UPE in 2-3 years showed a correlation coefficient more than 0.3 with serum cteatinine and e GFR.The pooled MCIDs of 2-3 years24h UPE were-1027mg/24h for the entire sample,-680mg/24h for subgroup of CKD G2,and-1027mg/24h for subgroup of non-nephrotic proteinuria.Preliminary MCIDsPreliminary MCIDs of 1-2 years 24h UPE were-1045mg/24h for the entire sample,-1552mg/24h and-953mg/24h subgroups of CKD stage 1 and 2,and-953mg/24h for subgroup of non-nephrotic proteinuria.Preliminary MCIDs of 2-3 years24h UPE were-1348mg/24h for the entire sample,-2474mg/24h and-1259mg/24h for subgroups of CKD stage 1 and 2,and-1356mg/24h for subgroup of non-nephrotic proteinuria.MDCs estimated by distribution-based methodsMDCs of 3-6 months 24h UPE were 189mg/24h for the entire sample,294mg/24h,361mg/24h,274mg/24h and 74mg/24h for subgroups of CKD stage 1,stage 2,nephrotic and non-nephrotic proteinuria.MDCs of 6 months to 1 year 24h UPE were 116mg/24h for the entire sample,201mg/24h,140mg/24h,257mg/24h and61mg/24h for subgroups of CKD stage 1,stage 2,nephrotic and non-nephrotic proteinuria.MDCs of 1-2 years 24h UPE were 99mg/24h for the entire sample,179mg/24h,117mg/24h,222mg/24h and 43mg/24h for subgroups of CKD stage 1,stage 2,nephrotic and non-nephrotic proteinuria.MDCs of 2-3 years 24h UPE were 109mg/24h for the entire sample,136mg/24h,148mg/24h,471mg/24h and118mg/24h for subgroups of CKD CKD stage 1,stage 2,nephrotic and non-nephrotic proteinuria.MDCs of 24h UPE for the entire sample ranged from 99to 189mg/24h.In order to avoid the false positive results,the final MDC was determined to be 189mg/24h.The final MCIDsPreliminary MCIDs were all greater than MDCs.That is,the reduction in proteinuria was real change but not causing by measurement error.Trials included in Meta analysis were evaluated effects of CHM in the short term(2-3 months),which were not consistent with the sample of this study,because the study excluded DKD patients who had a follow up duration less than 3 months.Consequently,MCIDs estimated by anchor-based methods were used to determine the final MCIDs.The absolute value of MCID of 1-3 years24h UPE ranged from 1045 to 1348mg/24h.Giving to the selection of the anchors,results of the MCID was not necessary to be nimimun,1045mg/24h(relative value 52%)was taken as MCID.In order to improve applicability,rounded to 1000mg/24h(relative change value 50%)was used as the final MCID.ConclusionMDC of 24h UPE in patients with T2DM complicated by early stages of DKD(CKD stage 1-2)after receiving treatment of CHM is 189mg/24h.Namely,a change of 24h UPE more than 189mg/24h is a real difference but not caused by a measurement error.The absolute and relative value of MCID of 24h UPE in this population who are receiving CHM therapy for 1-3 years are 1000mg/24h and 50%.That is to say,a reduction of 24h UPE more than 1000mg/24h or 50%reduction from baseline is considering as clinical significance.However,since the anchors used in this study were clinically meaningful endpoints or their surrogate endpoints for kidney diseases,they can distinguish patients who have progressed and who have not progressed,but cannot identify patients from mild deterioration and/or improvement and no change.In this case,the anchor-based estimated MCID was not necessary to be the minimal difference.Therefore,it should be cautious when applying the MCID of 24h UPE estimated in this study.