Introduction: Mayo clinic classification (MCC) has been proposed in patients with autosomal dominant polycystic kidney disease (ADPKD) to identify who may experience a rapid decline of renal function. Our aim was to validate this predictive model in a population from southern Spain. Methods: ADPKD patients with measurements of height-adjusted total kidney volume (HtTKV) and baseline estimated glomerular filtration rate (eGFR) >30 mL/min/1.73 m2 were selected. Last eGFR was estimated with Mayo Clinic (MC) equation and bias and accuracy were studied. We also analyzed predictive capacity of MCC classes using survival analysis and Cox regression models. Results: We included 134 patients with a mean follow-up of 82 months. While baseline eGFR was not different between classes, last eGFR decreased significantly with them. eGFR variation rate was different according to the MCC class with a more rapid decline in 1C, 1D, and 1E classes. Final eGFR predicted was not significantly different from the real one, with an absolute bias of 0.6 ± 17.0 mL/min/1.73 m2. P10 accuracy was low ranging from 37.5 to 59.5% in classes 1C, 1D, and 1E. Using MC equation, the rate of eGFR decline was underestimated in 1C, 1D, and 1E classes. Cox regression analysis showed that MCC class is a predictor of renal survival after adjusting with baseline eGFR, age, sex, and HtTKV, with 1D and 1E classes having the worst prognosis. Conclusion: MCC classification is able to identify patients who will undergo a more rapid decline of renal function in a Spanish population. Prediction of future eGFR with MC equation is acceptable as a group, although it shows a loss of accuracy considering individual values. The rate of eGFR decline calculated using MC equation can underestimate the real rate presented by patients of 1C, 1D, and 1E classes.

Relationship between cystic compartment growth and deterioration of glomerular filtration rate (GFR) in autosomal dominant polycystic kidney disease (ADPKD) was suggested many years ago by Franz and Reubi [1]. They stated that ADPKD evolved in 2 phases: an initial phase of stability in renal function and a second phase with a rapid drop of GFR. They postulated that a mathematical relationship could be established between both variables. Fick-Brosnahan et al. [2] observed with ultrasound that total kidney volume (TKV) estimated with ellipsoid formula correlated with GFR and that there was an inverse relationship with decline of renal function. This connection was clarified with the advent of computerized tomography (CT) and especially with magnetic resonance imaging (MRI), showing that reduction of kidney tissue precedes in years the GFR decline [3-5]. Moreover, it has been shown that TKV growth is mainly due to cystic compartment [6, 7], which correlates better with the GFR decline slope than TKV [8, 9]. For this reason, the Food and Drug Administration and the European Medicines Agency have recognized TKV as an early predictor to select patients in future trials [10, 11].

Follow-up of CRISP cohort has shown that kidney growth rate is quite stable over years [6] and that patients who undergo a faster GFR decline were those with higher kidney growth rate [8, 12]. Starting from this idea, Irazabal et al. [13, 14] built a predictive model for future GFR using a single TKV measurement and classifying patients into 5 classes according to growth rate (<1.5%, 1.5–3%, 3–4.5%, 4.5–6%, and >6% annually), observing different progression rates and different renal survivals between classes. This classification was validated in HALT-PKD [14, 15] and TEMPO [16] studies, showing its utility as early marker of evolution in ADPKD patients [16]. As a consequence, Mayo clinic classification (MCC) has been considered as a useful tool to identify rapid progressors in current guidelines [17-19].

MCC was developed in a population of ADPKD patients under follow-up at the Mayo Clinic (MC) and was validated with data from CRISP cohort, but it has not been validated in other population outside the USA, with different epidemiological characteristics which might alter its efficiency. Our aim was to analyze the usefulness of MCC predictive model in a population with ADPKD from the southern of Spain and to ascertain if it also has a long-term prognostic significance.

We included patients who had ADPKD based on family history, radiological findings, and/or genetic analysis. Patients were under follow-up as outpatients at Nephrology Department of “Virgen de las Nieves” University Hospital in Granada and at “Ciudad de Jaén” University Hospital in Jaén, both in the south of Spain. We selected patients who had a TKV measurement performed by CT or MRI, with an estimated eGFR with CKD-EPI equation >30 mL/min/1.73 m2 and with a minimum follow-up of 12 months after radiological measurement. We collected the last available eGFR at the end of follow-up or the closest to starting dialysis or before undergoing an anticipated kidney transplant.

TKV estimation was performed by stereological method in all people [20]. RMI was performed with T1-echo weighted sequences in-phase and opposite-phase gradient in the axial plane, T2-weigthted single-shot fast spin-echo (SSFSE/HASTE) sequence in axial and coronol axis, and T2-sequences with fast-relaxation fast spin-echo sequence with respiratory synchronization in axial plane. The MRI slice thickness was 6–9 mm and it is used Advantage Windows 4.6 from GE and PACS from Carestream 12.1.5 program for TKV calculation. CT images in axial axis were processed with Carestream 12.1.5 program using a slice thickness of 2–5 mm. TKV was adjusted by height (HtTKV) and patients were classified according to growth rate in MCC categories according to Irazabal et al. [13, 14].

To check whether MCC model could be reproduced with our data, we built a predictive model of eGFR at the end of follow-up using a general linear model, considering the last eGFR before entry into renal replacement therapy or at end of follow-up, as a dependent variable. As independent variables, we always introduced age, sex, and baseline eGFR, adding HtTKV first and then growth rate or MCC class. Time of follow-up was introduced as an independent variable or combined as growth rate x time or as MCC class x time, as Irazabal et al. [13, 14] did in their equation. In a subgroup of patients, we repeated this analysis including PKD1/PKD2 mutations as an independent variable. We presented eta squared coefficient for each predicting variable as a method for measuring the proportion of the total variance of the dependent variable that is associated with each independent variable, with similar meaning that determination coefficient in regression linear analysis.

For each patient, eGFR was calculated at the end of follow-up using predictive equation presented in Irazabal et al. [13, 14] publication, including the significant and nonsignificant variables, as authors did on MC website (https://www.mayo.edu/research/documents/pkd-center-adpkd-classification/doc-20094754). Absolute bias of this eGFR estimate was calculated as the difference between value predicted by MC equation and real eGFR value at the end of follow-up. Relative bias was obtained by dividing absolute bias by real eGFR and multiplying it by 100. Accuracy was estimated by calculating percentages of patients whose final eGFR differed up to a maximum of 10% (P10) or 30% (P30) from real eGFR.

We performed a survival analysis with MCC classification as predictor using Kaplan-Meier technique, considering as an event the entry in renal replacement therapy or presenting a final eGFR <10 mL/min/1.73 m2. Later, we performed a Cox regression analysis with MCC class as a categorical predictor and adjusting the model by baseline eGFR, age, sex, and HtTKV.

Statistical analysis was performed with SPSS v23 statistical package. Significant results were considered when p < 0.05. Quantitative variables were usually shown as mean ± standard deviation. Analysis of association between qualitative variables was performed using Pearson’s χ2 test. HtTKV was always included in predictive models as natural logarithm. Comparison between 2 groups was made with Mann-Whitney test and paired comparisons with Wilcoxon test. For comparisons between several groups, Kruskal-Wallis test was used. Correlation analysis was performed using Spearman rho coefficient.

We included 134 patients, 72 from “Virgen de las Nieves” Universitary Hospital and 62 from “Ciudad de Jaén” Universitary Hospital. In Table 1 we summarize characteristics of population. Mean age was 45 years and median of follow-up was 77 months, somewhat higher in women. Serum urea and creatinine levels were higher in men, with significantly higher eGFR in women. A genetic study was carried out in 100 patients, PKD1 being the most frequently mutation found, especially in men.

Table 1.

General data, TKV, mutation identified, and distribution according to MC classification in the population included in the study

General data, TKV, mutation identified, and distribution according to MC classification in the population included in the study
General data, TKV, mutation identified, and distribution according to MC classification in the population included in the study

HtTKV distribution versus age is shown in Figure 1. TKV and HtTKV were clearly higher in men. According to MCC, 3.7% belonged to class 1A, 30.6% to class 1B, 31.3% to class 1C, 22.4% to class 1D, and 11.9% to class 1E, which means that 67.2% of patients were in categories of high risk of progression: 1C, 1D, or 1E. Seventy-three percent of men were included in these classes, compared to 59.5% of women (p = 0.093, Pearson χ2).

Fig. 1.

Distribution of HtTKV and age of patients included in the study. HtTKV is represented with a natural logarithmic scale. Each group of MC classification is also shown in different colors representing HtTKV annual growth rate (1A <1.5%, 1B ≥1.5% & <3%, 1C ≥3% & <4.5%, 1D ≥4.5% & <6% and 1E ≥6%). HtTKV, height-adjusted total kidney volume; MC, Mayo Clinic.

Fig. 1.

Distribution of HtTKV and age of patients included in the study. HtTKV is represented with a natural logarithmic scale. Each group of MC classification is also shown in different colors representing HtTKV annual growth rate (1A <1.5%, 1B ≥1.5% & <3%, 1C ≥3% & <4.5%, 1D ≥4.5% & <6% and 1E ≥6%). HtTKV, height-adjusted total kidney volume; MC, Mayo Clinic.

Close modal

Table 2 summarizes characteristics according to MCC classes. HtTKV and kidney growth rate showed a progressive and significant increase from class 1A to class 1E, as well as a decreasing age and higher proportion of males. There was a significantly higher proportion of PKD1 mutations in classes 1D and 1E. We did not observe differences in baseline serum creatinine levels and eGFR among MCC classes. However, serum creatinine levels at the end of follow-up increased from class 1A to 1E but without differences in eGFR.

Table 2.

Distribution of the variables included in the study according to MCC

Distribution of the variables included in the study according to MCC
Distribution of the variables included in the study according to MCC

eGFR decline showed significantly higher rates from class 1A to class 1E. In Figure 2, we can observe the significant correlation found between HtTKV growth rate and rate of eGFR decline observed at the end of follow-up.

Fig. 2.

Annual growth rate of HtTKV and variation of eGFR between the last follow-up and the baseline eGFR corrected by time of follow-up. Annual rate of renal deterioration was negatively correlated with annual growth rate of HtTKV. HtTKV, height-adjusted total kidney volume; MC, Mayo Clinic; eGFR, estimated glomerular filtration rate.

Fig. 2.

Annual growth rate of HtTKV and variation of eGFR between the last follow-up and the baseline eGFR corrected by time of follow-up. Annual rate of renal deterioration was negatively correlated with annual growth rate of HtTKV. HtTKV, height-adjusted total kidney volume; MC, Mayo Clinic; eGFR, estimated glomerular filtration rate.

Close modal

MCC Class as a Predictor of Future eGFR

Using general linear model (MLG), we built predictive models for final eGFR that included different combinations of variables that are presented in Table 3 for comparative purposes. In the first model, we obtained that final eGFR was negatively related with higher HtTKV and that older age was associated with better final eGFR and with an average decrease of −2.89 mL/min per year. In the second model, age was not significant while a high growth rate was clearly related with a greater decrease in final eGFR. In the third model, the construct growth rate x time of follow-up showed that eGFR decreased −0.68% yearly, while kidney volume did not show significance. In model #4, we introduced MCC class separately of time of follow-up finding that 1C, 1D, and 1D classes were associated with progressive significant reductions in future eGFR. In model #5, we used the construct MCC class x years of follow-up as in Irazabal et al. [13, 14] equation, showing progressively increasing rate of eGFR annual reduction in MCC classes, ranging from −2.047 mL/min/year in 1B class to −3.919 mL/min/year in 1E class, validating the MC equation in our population.

Table 3.

Predictive models of eGFR at last follow-up using GLM

Predictive models of eGFR at last follow-up using GLM
Predictive models of eGFR at last follow-up using GLM

As it can be seen in Table 3, all models predicted around 76% of data variance, but the impact of each variable was very different. Baseline eGFR was the quantitatively more important predictor of future eGFR, with partial eta2 ranging from 55.2% in model #2 to 62.1% in model #5, equation similar to MC equation. The second more important variable was the time of follow-up with partial eta2 around 27.9–29.1% when it was introduced independently in the models. The impact of kidney volume was lower with an eta2 of 9.1% in model #1 and with 10.3% for growth rate in model #2. Mayo classes had low values of eta2 (3–7.9%) when they were considered independently of time in model #4, but their impact increased when they were introduced as the construct Mayo class x years of follow-up rising eta2 to 12.8 for 1B class to 27.9% for 1D class.

Patients with PKD1 mutation showed a higher and significant decline rate of eGFR than PKD2 (−4.70 ± 3.40 vs. −2.56 ± 3.64 mL/min/year; p = 0.026). In Table 4, we show an MLG model, including PKD mutation using PKD2 as reference and, as it can be seen, PKD1 mutation was associated with significant higher decline of eGFR in addition to the effect of MCC class.

Table 4.

Predictive model of eGFR at last follow-up using GLM, including PKD mutation in addition to MCC

Predictive model of eGFR at last follow-up using GLM, including PKD mutation in addition to MCC
Predictive model of eGFR at last follow-up using GLM, including PKD mutation in addition to MCC

Bias and Accuracy of Predictions Using MCC Class

Final eGFR was 55 ± 33 mL/min/1.73 m2 and the value estimated by MC equation was 56 ± 27 mL/min/1.73 m2 (p = 0.677, Wilcoxon test). Absolute bias was 0.6 ± 17.0 mL/min/1.73 m2 and relative bias was 35.1 ± 109%. In Table 5 we show bias according to MCC. While absolute bias was low relative bias increased from 1A class to 1E class. Moreover, P10 and P30 accuracy values progressively decreased toward 1E class.

Table 5.

Comparison between GFR in last follow-up versus predicted with MC equation and between GFR variations from baseline when MC equation is used

Comparison between GFR in last follow-up versus predicted with MC equation and between GFR variations from baseline when MC equation is used
Comparison between GFR in last follow-up versus predicted with MC equation and between GFR variations from baseline when MC equation is used

Figure 3a shows final eGF calculated with MC equation against real eGFR values. All MCC classes were distributed covering the total range of both variables. Although correlation was good (r = 0.87, p < 0.001), bias shown individually was large and highly variable (Fig. 3b), with most of cases contained in an interval of −30 to +40 mL/min/1.73 m2. Furthermore, a clear correlation can be observed between real final eGFR and bias committed (r = −0.53), with a clear tendency to underestimate eGFR values in patients with higher final eGFR. We did not observe any relationship with baseline eGFR.

Fig. 3.

a Predicted eGFR with the MC equation was positively well correlated with eGFR presented in the last follow-up. b Absolute bias between eGFR predicted with MC equation versus last eGFR was negatively correlated with values of eGFR in the last follow-up. MC, Mayo clinic; eGFR, estimated glomerular filtration rate.

Fig. 3.

a Predicted eGFR with the MC equation was positively well correlated with eGFR presented in the last follow-up. b Absolute bias between eGFR predicted with MC equation versus last eGFR was negatively correlated with values of eGFR in the last follow-up. MC, Mayo clinic; eGFR, estimated glomerular filtration rate.

Close modal

We calculated the predicted eGFR decline rate by subtracting final eGFR using MC equation from baseline eGFR (Table 4). Globally, predicted eGFR decline rate was −3.3 ± 1.2 mL/min/1.73 m2/year and observed eGFR decline was −3.7 ± 3.4 mL/min/1.73 m2/year, difference not significant. We can observe that when passing from class 1A to 1E, estimated eGFR decline rate was progressively higher, although not significantly. These eGFR decline values were lower than real rates shown by patients in 1C, 1D, and 1E classes.

Figure 4 shows the predicted eGFR decline with MC equation versus real eGFR decline considering eGFR at the end of follow-up. We observed poor correlation (r = 0.33, p < 0.001), a great dispersion of data, and a narrower interval of values. MC equation predicted a deterioration of eGFR in the vast majority of patients and in all classes, especially in 1C, 1D, and 1E classes.

Fig. 4.

Variation of predicted eGFR using MC equation versus baseline eGFR showed a significant but poor correlation with actual variation rate of eGFR. MC, Mayo clinic; eGFR, estimated glomerular filtration rate.

Fig. 4.

Variation of predicted eGFR using MC equation versus baseline eGFR showed a significant but poor correlation with actual variation rate of eGFR. MC, Mayo clinic; eGFR, estimated glomerular filtration rate.

Close modal

At the end of follow-up, 22 (16.4%) patients had an eGFR <10 mL/min/1.73 m2 or were included in renal replacement therapy. None was predicted by the MC equation (sensitivity). 112 patients did not have this eGFR at the end of follow-up and 110 (98.2%) were classified correctly with MC prediction (specificity). The positive predictive value of MC equation was 0% (0/22) and the negative predictive value was 83.3% (110/132).

Renal Survival Analysis

Distribution by MC classes and mean age of patients who presented a final eGFR <10 mL/min/1.73 m2 or were included in renal replacement therapy were as follows: class 1A 0 patient, class 1B 4 patients and 73 years, class 1C 7 patients and 55 years, class 1D 6 patients and 54 years, and class 1E 5 patients and 42 years (χ2 test not significant and Kruskal-Wallis test p = 0.002). Figure 5 shows the renal survival curves (Kaplan-Meier technique) according to age where it can be seen that 1E and 1D classes lost renal function before 50 and 60 years, respectively, being patients with the earliest evolution to end-stage renal failure. Sixteen patients (21.6%) with PKD1 mutations presented a final event compared to 0 with PKD2 mutations (p = 0.063).

Fig. 5.

Renal survival curves using Kaplan-Meier analysis showing probability of not presenting an eGFR <10 mL/min/1.73 m2 according to age, separating by class of MC classification. Each class is clearly separated from others undergoing a progressive reduction in age from 1A to 1E classes. MC, Mayo clinic; eGFR, estimated glomerular filtration rate.

Fig. 5.

Renal survival curves using Kaplan-Meier analysis showing probability of not presenting an eGFR <10 mL/min/1.73 m2 according to age, separating by class of MC classification. Each class is clearly separated from others undergoing a progressive reduction in age from 1A to 1E classes. MC, Mayo clinic; eGFR, estimated glomerular filtration rate.

Close modal

Figure 6a shows renal survival curve of patients grouped according to MCC using the Kaplan-Meier technique, considering 1A + 1B classes as reference. The 1C and 1D classes were clearly separated from 1A + 1B and 1E classes, but not between them (log-rank p = 0.067). The 10-year survival for each MCC class was: 1A + 1B 84.4%, 1C 71.8%, 1D 66.7%, and 1E 44.4%.

Fig. 6.

a Renal survival separating by class of MC classification. We observed a clear decreasing renal survival for patients in 1E class and in 1C and 1D class from 1A + 1B classes, but log-rank test was not significant. b MC classes were significant predictors of renal survival when a regression Cox model was employed (curves were corrected by baseline eGFR, age, and sex and the natural logarithm of HtTKV). MC, Mayo clinic; eGFR, estimated glomerular filtration rate; HtTKV, height-adjusted total kidney volume.

Fig. 6.

a Renal survival separating by class of MC classification. We observed a clear decreasing renal survival for patients in 1E class and in 1C and 1D class from 1A + 1B classes, but log-rank test was not significant. b MC classes were significant predictors of renal survival when a regression Cox model was employed (curves were corrected by baseline eGFR, age, and sex and the natural logarithm of HtTKV). MC, Mayo clinic; eGFR, estimated glomerular filtration rate; HtTKV, height-adjusted total kidney volume.

Close modal

Since renal survival depends on baseline eGFR, we performed a Cox regression analysis adjusting by baseline eGFR, MC class, age, sex, and HtTKV. Table 6 shows only significant variables in 2 models. In model 1, MC class was included as in the work of Irazabal’s group [21] with numerical values for each class (1A + 1B = 1; 1C = 2, 1D = 3, and 1E = 4), obtaining a significant hazard ratio (HR) of 2,245. In the second model, we included MCC class as a categorical variable, obtaining as HR for 1C class 3,064, for 1D class 3917, and for 1E class 15,406, the latter 2 being significant. In both models, baseline eGFR was a significant variable, entering as a protective factor with HR around 0.93. Survival curves for each MCC class are shown in Figure 6b in which a progressively decreasing survival can be observed from classes 1A + 1B to 1E. We did not include mutations in the model because insufficient sample for PKD2 mutation.

Table 6.

Survival analysis using Cox regression analysis using Mayo classification as a quantitative variable (model #1) or as a qualitative variable (model #2), separating each class of Mayo classification (1A + 1B class is the reference category in both models)

Survival analysis using Cox regression analysis using Mayo classification as a quantitative variable (model #1) or as a qualitative variable (model #2), separating each class of Mayo classification (1A + 1B class is the reference category in both models)
Survival analysis using Cox regression analysis using Mayo classification as a quantitative variable (model #1) or as a qualitative variable (model #2), separating each class of Mayo classification (1A + 1B class is the reference category in both models)

Our objective was to evaluate whether MCC classification could be applied to a Spanish population of ADPKD patients, with evolutionary characteristics potentially different. We have reproduced MCC model using our data, confirming that MCC is a predictor of future eGFR, explaining around 76% of total variance. End of follow-up eGFR in our study progressively decreased from class 1B to 1E, with a significantly higher eGFR decline rates in classes 1C, 1D, and 1E, which support the predictive nature of MC classification. The fact that MC equation is adjusted to baseline eGFR, supports the observation that kidney growth rate is an early marker of future eGFR deterioration in ADPKD patients.

Although the model built by Irazabal et al. [13, 14] is significant, it is important to say that baseline eGFR was the most important variable in prediction in this model, assuming 55–57% of variance in our data. In the data of Irazabal et al. [13, 14], MC class only represented a 3% of improvement in explained variance. Thus, the variable most correlated with future eGFR is the proper baseline eGFR. MCC has a small value from a quantitative point of view, although very important from a pathogenic point of view. Moreover, authors have kept included on MC Web site some variables that are not significant. We think that it would be more convenient to delete them and publish the coefficients only for the significant variables.

In univariate analysis, we observed a higher rate of eGFR decline in patients with PKD1 mutation with a lower renal survival when PKD2 mutation was present, but not significantly due to a low number of patients. When mutations were included in the MLG model, we found a significant deleterious effect for PKD1 mutation in addition to MCC, while this was not found in a previous study [21]. PKD1 and PKD2 mutations show similar cyst growth rates but different kidney growth rate that has been related to a greater number of cysts in PKD1 mutation [22].

Future eGFR estimates with MC equation seemed very small if we look the mean bias, but large if we center our attention in standard deviation or in relative bias that rised up to 63.7% in 1E class. This means a lack of accuracy in the individual estimates. The same was observed by Irazabal et al. [13, 14] both in internal and external validation groups. Ability to predict a GFR <45 mL/min/1.73 m2 starting from a eGFR >80 mL/min/1.73 m2 improved from 17.4% to 47.8% when MC classes were included. On the contrary, when starting eGFR was <60 mL/min/1.73 m2, predictive capacity mildly improved with MCC classes (from 95.1 to 96.7%). In our data, accuracy was worse in classes 1C, 1D, and 1E.

eGFR decline rates were higher in our data than that predicted by MC equation, especially in classes 1C, 1D, and 1E. There may be several explanations for this discrepancy. GFR deterioration in ADPKD patients is curvilinear and not linear as MCC model proposes, which will cause a greater final drop in eGFR and higher bias in 1D and 1E classes at longer follow-up times [23]. On the other hand, it is possible that in our cohort patients with a TKV measurement have undergone a worse evolution because we have not performed systematically TKV measurement to all patients included in follow-up as it was done in CRISP study.

In the publication of Irazabal et al. [13, 14], curves of renal survival with Kaplan-Meier analysis showed a perfect separation of MCC classes in internal validation group but more later in external validation, without total separation of classes. We observed that 1C and 1D classes showed similar survivals without achieving separation until the end of follow-up. It must be remembered that Kaplan-Meier survival analysis only analyzes strata of one variable, so results can be greatly influenced by baseline eGFR values which were similar in both groups in our data and dissimilar in the work of Irazabal et al. [13, 14].

When we used Cox regression analysis adjusting by baseline eGFR, survival curves showed a total separation of survival for all classes, supporting validity of MC classification in our population. Irazabal et al. [13, 14] also performed a multivariate analysis and found a significant HR for MC classes in both MC and CRISP cohorts. In their analysis, they represented MCC class as a quantitative variable that supposes equidistance between each category. It is not really known whether kidney growth rate has a linear relationship with renal survival. Therefore, we preferred considering a categorical variable to represent each MCC class, and thus we estimated risk of each class independently. In this way we showed that the risk of entering in renal replacement therapy or presenting an eGFR <10 mL/min/1.73 m2 was very different in each class and especially important for class 1E.

Our work has some limitations that must be pointed out. Our sample size was smaller than that used by Irazabal et al. [13, 14] to build their model. However, it is enough to validate it and test the basic hypothesis of the MC model, namely, kidney growth rate influences the progressive deterioration of glomerular filtration, which was our aim. In fact, the cohorts of validation used by Irazabal et al. [13, 14] are only slightly larger than ours. We included TKV measurements performed with CT and MRI in our study as did Irazabal et al. [13, 14]. They showed that the difference between both techniques was small, from −3.1 to +3.6%, and lower than the annual growth rate, considering that this is not a limitation. On the other hand, we performed TKV measurements using segmentation while Irazabal et al. [13, 14] preferred the ellipsoid method in the building of their model because of the very good correlation that both methods have. Although there might be a small difference between measurements, both will be probably included within the same MCC class causing little error. We used GFR estimated by the CKD-EPI equation as Irazabal et al. [13, 14] did, and we do not know if using actual GFR measured with other methods would result in a model with greater or lesser precision. Proportion of patients that developed end-stage renal failure in our cohort was low (21.6%) after a mean follow-up time of 72–91 months, which is a low number of events to perform a renal survival study. However, this low proportion also was present in the cohort of Irazabal et al. [13, 14] and in the CRISP cohort. Even so, MC classification showed a significant predictive role when it was adjusted by baseline eGFR in our data and in the publication of Irazabal et al. [13, 14], validating the model despite this important limitation.

Therefore, we show that MC classification is valid in Spanish population of patients with ADPKD. MC equation can be used to estimate future eGFR and to obtain the rate of eGFR decline that it will undergo in subsequent years, but only for population groups. At individual level, these predictions should be taken with great caution since the individual error committed can be very high, especially in patients with a good baseline eGFR. We believe that it is necessary to continue publishing results in series with a longer evolution time and with a higher proportion of patients entering in renal replacement therapy, in order to make more precise predictions.

Oral consent was obtained from all patients to include anonymized data in this study. All data were collected with approval of Local Ethical Board and with adherence to the International Conference of Harmonization – Good Clinical Practice.

Authors have no conflict of interest to declare. We declare that the results presented in manuscript have not been published previously in whole or part, except in abstract format.

We have not received any financial support for this research.

Research idea and study design: F.J.B.U.; data acquisition: F.J.B.U., R.E.R., E.M.G., A.P.M., C.M.D., A.M.D., A.I.M.G., and J.A.B.S.; data analysis/interpretation: F.J.B.U., R.E.R., and J.A.B.S.; statistical analysis: F.J.B.U. Each author contributed significantly in the content during manuscript drafting and revision. All authors approved the final version of the manuscript.

1.
Franz
KA
,
Reubi
FC
.
Rate of functional deterioration in polycystic kidney disease
.
Kidney Int
.
1983
;
23
(
3
):
526
9
. .
2.
Fick-Brosnahan
GM
,
Belz
MM
,
McFann
KK
,
Johnson
AM
,
Schrier
RW
.
Relationship between renal volume growth and renal function in autosomal dominant polycystic kidney disease: a longitudinal study
.
Am J Kidney Dis
.
2002
;
39
(
6
):
1127
34
. .
3.
Grantham
JJ
,
Mulamalla
S
,
Swenson-Fields
KI
.
Why kidneys fail in autosomal dominant polycystic kidney disease
.
Nat Rev Nephrol
.
2011
;
7
(
10
):
556
66
. .
4.
Grantham
JJ
,
Chapman
AB
,
Torres
VE
.
Volume progression in autosomal dominant polycystic kidney disease: the major factor determining clinical outcomes
.
Clin J Am Soc Nephrol
.
2006
;
1
(
1
):
148
57
. .
5.
Chapman
AB
,
Bost
JE
,
Torres
VE
,
Guay-Woodford
L
,
Bae
KT
,
Landsittel
D
,
Kidney volume and functional outcomes in autosomal dominant polycystic kidney disease
.
Clin J Am Soc Nephrol
.
2012
;
7
(
3
):
479
86
. .
6.
Grantham
JJ
,
Torres
VE
,
Chapman
AB
,
Guay-Woodford
LM
,
Bae
KT
,
King
BF
 Jr.
,
Volume progression in polycystic kidney disease
.
N Engl J Med
.
2006
;
354
(
20
):
2122
30
. .
7.
Grantham
JJ
,
Cook
LT
,
Torres
VE
,
Bost
JE
,
Chapman
AB
,
Harris
PC
,
Determinants of renal volume in autosomal-dominant polycystic kidney disease
.
Kidney Int
.
2008
;
73
:
108
16
. .
8.
King
BF
,
Reed
JE
,
Bergstralh
EJ
,
Sheedy
PF
 2nd
,
Torres
VE
.
Quantification and longitudinal trends of kidney, renal cyst, and renal parenchyma volumes in autosomal dominant polycystic kidney disease
.
J Am Soc Nephrol
.
2000
;
11
(
8
):
1505
11
. .
9.
Higashihara
E
,
Nutahara
K
,
Okegawa
T
,
Shishido
T
,
Tanbo
M
,
Kobayasi
K
,
Kidney volume and function in autosomal dominant polycystic kidney disease
.
Clin Exp Nephrol
.
2014
;
18
(
1
):
157
65
. .
10.
US Food and Drug Administration
.
Qualification of biomarker: total kidney volume in studies for treatment of autosomal dominant polycystic kidney disease
. Available from: http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM458483.pdf. Accessed 2016 Dec 30.
11.
European Medicines Agency
.
Qualification opinion. Total kidney volume (TKV) as a prognostic biomarker for use in clinical trials evaluating patients with autosomal dominant polycystic kidney disease (ADPKD)
. Available from: http://www.ema.europa.eu/docs/en_GB/document_library/Regulatory_and_procedural_guideline/2015/11/WC500196569.pdf. Accessed 2016 Dec 30.
12.
Yu
ASL
,
Shen
C
,
Landsittel
DP
,
Harris
PC
,
Torres
VE
,
Mrug
M
,
Baseline total kidney volume and the rate of kidney growth are associated with chronic kidney disease progression in autosomal dominant polycystic kidney disease
.
Kidney Int
.
2018
;
93
(
3
):
691
9
. .
13.
Irazabal
MV
,
Rangel
LJ
,
Bergstralh
EJ
,
Osborn
SL
,
Harmon
AJ
,
Sundsbak
JL
,
Imaging classification of autosomal dominant polycystic kidney disease: a simple model for selecting patients for clinical trials
.
J Am Soc Nephrol
.
2015 Jan
;
26
(
1
):
160
72
. .
14.
Irazabal
MV
,
Abebe
KZ
,
Bae
KT
,
Perrone
RD
,
Chapman
AB
,
Schrier
RW
,
Prognostic enrichment design in clinical trials for autosomal dominant polycystic kidney disease: the HALT-PKD clinical trial
.
Nephrol Dial Transplant
.
2017
;
32
(
11
):
1857
65
. .
15.
Brosnahan
GM
,
Abebe
KZ
,
Moore
CG
,
Rahbari-Oskoui
FF
,
Bae
KT
,
Grantham
JJ
,
Patterns of kidney function decline in autosomal dominant polycystic kidney disease: a post hoc analysis from the HALT-PKD trials
.
Am J Kidney Dis
.
2018
;
71
(
5
):
666
76
. .
16.
Irazabal
MV
,
Blais
JD
,
Perrone
RD
,
Gansevoort
RT
,
Chapman
AB
,
Devuyst
O
,
Prognostic enrichment design in clinical trials for autosomal dominant polycystic kidney disease: the TEMPO 3:4 clinical trial
.
Kidney Int Rep
.
2016
;
1
(
4
):
213
20
. .
17.
Chebib
FT
,
Perrone
RD
,
Chapman
AB
,
Dahl
NK
,
Harris
PC
,
Mrug
M
,
A practical guide for treatment of rapidly progressive ADPKD with tolvaptan
.
J Am Soc Nephrol
.
2018 Oct
;
29
(
10
):
2458
70
. .
18.
Müller
RU
,
Haas
CS
,
Sayer
JA
.
Practical approaches to the management of autosomal dominant polycystic kidney disease patients in the era of tolvaptan
.
Clin Kidney J
.
218
;
11
(
1
):
62
9
. .
19.
Edwards
ME
,
Chebib
FT
,
Irazabal
MV
,
Ofstie
TG
,
Bungum
LA
,
Metzger
AJ
,
Long-term administration of tolvaptan in autosomal dominant polycystic kidney disease
.
Clin J Am Soc Nephrol
.
2018 Aug 7
;
13
(
8
):
1153
61
. .
20.
Chapman
AB
,
Guay-Woodford
LM
,
Grantham
JJ
,
Torres
VE
,
Bae
KT
,
Baumgarten
DA
,
Consortium for radiologic imaging studies of polycystic kidney disease cohort. Renal structure in early autosomal-dominant polycystic kidney disease (ADPKD): the Consortium for Radiologic Imaging Studies of Polycystic Kidney Disease (CRISP) cohort
.
Kidney Int
.
2003 Sep
;
64
(
3
):
1035
45
.
21.
Yu
ASL
,
Shen
C
,
Landsittel
DP
,
Harris
PC
,
Torres
VE
,
Mrug
M
,
Baseline total kidney volume and the rate of kidney growth are associated with chronic kidney disease progression in autosomal dominant polycystic kidney disease
.
Kidney Int
.
2018
;
93
:
691
9
. .
22.
Harris
PC
,
Bae
KT
,
Rossetti
S
,
Torres
VE
,
Grantham
JJ
,
Chapman
AB
,
Cyst number but not the rate of cystic growth is associated with the mutated gene in autosomal dominant polycystic kidney disease
.
J Am Soc Nephrol
.
2006
;
17
:
3013
9
. .
23.
Yu
ASL
,
Shen
C
,
Landsittel
DP
,
Grantham
JJ
,
Cook
LT
,
Torres
VE
,
Consortium for radiologic imaging studies of polycystic kidney disease (CRISP). Long-term trajectory of kidney function in autosomal-dominant polycystic kidney disease
.
Kidney Int
.
2019
;
95
(
5
):
1253
61
.
Copyright / Drug Dosage / Disclaimer
Copyright: All rights reserved. No part of this publication may be translated into other languages, reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying, recording, microcopying, or by any information storage and retrieval system, without permission in writing from the publisher.
Drug Dosage: The authors and the publisher have exerted every effort to ensure that drug selection and dosage set forth in this text are in accord with current recommendations and practice at the time of publication. However, in view of ongoing research, changes in government regulations, and the constant flow of information relating to drug therapy and drug reactions, the reader is urged to check the package insert for each drug for any changes in indications and dosage and for added warnings and precautions. This is particularly important when the recommended agent is a new and/or infrequently employed drug.
Disclaimer: The statements, opinions and data contained in this publication are solely those of the individual authors and contributors and not of the publishers and the editor(s). The appearance of advertisements or/and product references in the publication is not a warranty, endorsement, or approval of the products or services advertised or of their effectiveness, quality or safety. The publisher and the editor(s) disclaim responsibility for any injury to persons or property resulting from any ideas, methods, instructions or products referred to in the content or advertisements.