Autosomal dominant polycystic kidney disease (ADPKD) is caused by mutations in two genes,
PKD1 and PKD2, which code for polycystin proteins that are involved in ciliary function of the renal epithelial cell ( ). 1 PKD1 is associated with earlier age to ESRD/death than PKD2 ( 2 , ). 3
Few longitudinal studies that have investigated the clinical course of renal disease in ADPKD have been reported. Hypertension studies usually have been cross-sectional, providing data on prevalence but not on incidence (
4 – ). Studies of renal function have been influenced by referral and ascertainment bias, involved multiple probands and small family size ( 6 7 , ), and provide few data on incidence of chronic kidney disease (CKD). Good information exists on age to dialysis/death, but, frequently, patients who were enrolled were not representative of the ADPKD population ( 8 5 , ). 8
Information on other risk factors for incident renal events in ADPKD is limited. The impact of gender is uncertain (
4 – 6 , 8 – ), gender of the parent from whom the ADPKD was inherited has been reported to be a risk factor ( 13 10 , ), phenotypic differences exist between families ( 13 5 , ), and a family history of essential hypertension may influence clinical outcomes ( 10 13 , ). Furthermore, parity may influence progression of CKD ( 14 ) and of ADPKD ( 15 ), and the use of the oral contraceptive pill (OCP) may be conducive to progression of CKD by increasing BP and filtration function and activating the renin angiotensin system ( 7 ). 16
Newfoundland, an island in the North Atlantic, is characterized by founder effects, large family size with family members settling near the core community, and little in- or out-migration since the founding migrations from Southeast Ireland and Southwest England in the late 18th and early 19th centuries. Investigation of large families with autosomal dominant disorders has occurred (
), facilitated by their eager participation in research and good access to the publicly funded Canadian health care system. In addition, exposure to a homogeneous environment diminishes the role of environmental factors as a potential confounder of clinical outcomes. 17
In 1981, we identified all patients who had ADPKD and attended nephrology/urology clinics in the province and ascertained family members who were at 50% risk for inheriting ADPKD. We subsequently reported the diagnostic utility of renal imaging and clinical outcomes in PKD1 and non-PKD1 (
2 , ), investigated the pathogenesis of hypertension ( 10 ), and collaborated to identify the genetic basis of cyst development in PKD2 ( 18 19 , ). This cohort has little ascertainment bias and has been followed for 22 yr at serial research clinics. In this population-based and family-derived cohort, we report the probability of development of renal events (hypertension requiring treatment, stage 3 CKD, ESRD, and death) in PKD1 and PKD2 and assess factors that may have an impact on these risks. 20 Materials and Methods
Eighteen families, identified through the patients who attended nephrology/urology clinics across Newfoundland in 1981, were investigated. Family members who were at 50% risk for inheriting ADPKD had a clinical evaluation and renal ultrasound performed and DNA sample taken.
To determine whether the families who were involved in this study were representative of the provincial population, we studied the cause of renal disease in all prevalent dialysis patients in 1987 and incident patients from 1987 to 1993 (
). Forty-six individuals had PKD as the cause of ESRD: 36 were already enrolled in our study, six did not have a family history of ADPKD, and the remaining four did not wish to participate in further studies ( 21 ). Subsequently, in 2003, we contacted all nephrologists and urologists who were practicing in the province to identify individuals who had received a diagnosis of ADPKD. Through the multiple referrals received, only one individual was unknown to the investigators, and upon review of the family pedigree, it was thought that this individual was a new mutation, because no other family history of ADPKD could be confirmed. From these two studies, we could conclude that although probands who entered the study were exposed to referral bias, the cohort enrolled in our study was representative of the population. 16
In 14 of 18 families, a disease-associated haplotype was identified at either the PKD1 or PKD2 locus. The remaining four (22%) families were small and uninformative from a genetic perspective. Family members were considered to have inherited ADPKD gene when (
1) they had the PKD disease–associated haplotype or ( 2) they had renal cysts according to criteria of Ravine et al. ( ). They were considered unaffected when ( 22 1) they did not carry the PKD-associated haplotype or ( 2) they did not have renal cysts and were older than 30 yr. In the 14 families, some going back seven generations, 613 individuals were at 50% risk for inheriting ADPKD in 147 sibships. To diminish ascertainment bias, we studied only members of well-ascertained sibships in whom at least 50% of siblings had known ADPKD disease status. This comprised 371 (61%) family members from 62 well-ascertained sibships who were at 50% risk for inheriting PKD. Within this group, 194 (52%) had ADPKD, 129 (35%) did not have ADPKD, nine (2%) were obligate carriers, nine (2%) had an equivocal diagnosis, and 43 (12%) had unknown status. It is unclear why there was an unequal distribution of affected and unaffected, but it is related to ascertainment in the PKD1 family (P17), in which 57 were affected and 16 were not ( Table 1).
Four research clinics were held at 6-yr intervals (1982, 1988, 1994, and 2000), at which time family members were interviewed and examined; medical records were reviewed; and data were abstracted concerning smoking habits, body mass index (BMI), angiotensin-converting enzyme (ACE) inhibitor and statin use, hypertension diagnosis and treatment, serum creatinine levels, treatment of ESRD, and cause of death. BP and serum creatinine were measured. GFR was estimated using the Modification of Diet in Renal Disease formula (
). Age at hypertension diagnosis and treatment was determined; the prevalence and the incidence of stage 3 CKD were assessed; and age at ESRD, death, or last follow-up was recorded. Because age at hypertension diagnosis was followed closely by treatment, only the latter is discussed here. 23
The following risk factors for each renal event were studied: Genotype (PKD1
versus PKD2), gender, gender of parent who transmitted ADPKD, and family history of essential hypertension (presence of hypertension treated before the age of 60 in an unaffected first- or second-degree relative) ( ). Outcomes in families with >10 affected individuals were compared. In women, parity (zero to two 14 versus three or more children) and use of the OCP (≥1 versus no or <1 yr of use) were assessed. Statistical Analyses
PKD1 and PKD2 family members were analyzed separately. In addition, PKD1 families P6, P7, P10, and P18 were compared with P17. The two PKD2 families with the same mutation were compared with PKD2 families with independent mutations. Time to each event was measured using Kaplan-Meier methods (
). The impact of each risk factor was assessed using Cox regression ( 29 ). Multivariate modeling was used where indicated. 30
For each outcome of interest as described, the Cox procedure was extended to test the independent role of variables with updated values over time (BMI, use of ACE inhibitors and statins, and smoking habits and proteinuria). In all cases, model specification and overall fit were checked by re-estimation; by formal tests based on Schoenfeld, Martingale, and Cox-Snell residuals; and testing the interaction with time of the variables in the model. Influence analysis was conducted on the basis of efficient score residuals. Within sensitivity analysis results, consistency was verified excluding and including in either group the two individuals who were found to be transheterozygous for both PKD1 and PKD2. All analyses were performed using STATA 9.1 SE (StataCorp, College Station, TX). Because of missing data, the members who were included in the analysis of each renal event are different.
Nine (64%) of 14 informative families had PKD1, four (29%) had PKD2, and one had bilineal inheritance of both PKD1 and PKD2, previously described by Pei
et al. ( ). The number of family members from well-ascertained PKD1 sibships in whom cysts were identified was 136 and from PKD2 sibships was 60. 24 Table 1 describes each family by logarithm of odds score for disease associated haplotype, number with and without cysts, number with equivocal ultrasound results and obligate carriers. Mutations
The largest family, P17, has a low logarithm of odds score (0.86) as a result of two independent recombinations. The
PKD1 mutation was 11587 del. 1 bp in the nonduplicated region of exon 40 (AA 3792 FS). This was not present in the other eight PKD1 families.
Two PKD2 families, P3 and P16, had the same disease haplotype and mutation (C→T1390, R464 X in exon 6). A third family (P13) from the same locality had the same disease haplotype but did not have the same mutation. The PKD2 mutation in P10 was 2152 del. A; L736 X in exon 11 (
). 24 Hypertension
The age to onset of hypertension treatment in those with and without cysts in PKD1 and PKD2 is presented in
Table 2. The median age to hypertension treatment was 46 yr in PKD1 and 51 yr in PKD2. The youngest age at hypertension treatment was 18 yr in PKD1 and 20 yr in PKD2. In PKD1, the relative risk for hypertension treatment was 9.8 (95% confidence interval [CI] 3.5 to 27.8) in those with cysts compared with the unaffected family members. In PKD2, the relative risk was 4.1 (95% CI 1.6 to 10.8).
No difference in incidence of hypertension treatment was observed in PKD1 when analyzed by gender, parent who transmitted PKD, family history of essential hypertension, parity, and use of OCP (
Table 3). In the five large PKD1 families with >10 affected family members, there was variability in the age to hypertension onset, with median ranging from 35 yr in P6 to 62 yr in P10 ( Table 4). Age to onset of hypertension in P6 was significantly earlier compared with that in P17 (hazard ratio [HR] 4.08; 95% CI 1.53 to 10.8).
In the two PKD2 families with the known mutation, the median age to onset of hypertension treatment was 48 yr, and for the three combined PKD2 families with independent mutations, the median age was 53 yr. No risk factors were identified for the PKD2 group (
Table 3). CKD Stage 3
Serum creatinine was performed in 118 (87%) of 136 PKD1 cases at mean age of 27 ± 14.3 SD yr, and GFR was estimated (
Figure 1A). The prevalence of stage 3 CKD at first measurement was 20% ( n = 24). Seventeen of these cases had serial serum creatinine levels. The mean rate of progression of CKD was 3.0 ± 2.1 ml/min per yr. In 94 cases with estimated GFR >60 ml/min at baseline, the median age of de novo development of stage 3 CKD was 50 yr ( Table 5). The youngest age of diagnosis of CKD stage 3 was 35 yr.
Serum creatinine was performed in 48 (80%) of 60 PKD2 cases at a median age of 40 ± 15 SD yr (
Figure 1B). The prevalence of stage 3 CKD was 38% ( n = 18). The mean rate of progression in this group ( n = 15) was 3.5 ± 2.1 ml/min per yr. In 30 cases with estimated GFR >60 ml/min, the median age to development of CKD stage 3 was 66 yr ( Table 5). The youngest age at diagnosis of CKD stage 3 was 48 yr. The median age to diagnosis of CKD stage 3 in the incident cohort in PKD2 families with the same mutation was 66 yr and for the three combined PKD2 families with unknown mutations was 65 yr. ESRD
The median age to onset of ESRD for PKD1 was 53 yr (
Table 5). In PKD2, only 21% ( n = 6) developed ESRD by age 70 yr. The youngest age at ESRD in PKD1 was 30 yr and in PKD2 was 42 yr.
There was no difference in the incidence of ESRD in PKD1 when analyzed by gender, parent who transmitted PKD1, family history of essential hypertension, parity, or taking the OCP (
Table 6). In the families with >10 affected members, P18 had a lower median age to onset of ESRD than the other large families ( Table 6), which was significantly earlier then in P17 (HR 13.42; 95% CI 3.77 to 47.81).
In family P10, in which inheritance of both PKD1 and PKD2 occurred, two members inherited both PKD1 and PKD2. Both developed ESRD at ages 48 and 52 yr (
). 24 Mortality
In PKD1, median age to death in those with cysts was 67 yr and in PKD2 was 71 yr (
Table 7). In PKD1, cause of death was known in 11 of 14 cases: Uremia in four (36%), cerebral hemorrhage in two (18%), ruptured cerebral aneurysm in two (18%), cancer in two (18%), and other causes in one (10%). In PKD2, cause of death was known in 16 of 17 cases: Uremia in two (13%), cerebral hemorrhage in four (25%), ruptured cerebral aneurysm in two (13%), cardiac disease in one (6%), cancer in four (25%), and other causes in three (19%). Impact of ESRD Therapy
In PKD1, 24 patients received dialysis or transplantation, seven of whom died. The mean survival from time of initiation of therapy after ESRD to death was 15.2 yr.
Potential Confounding Factors
Considering the variable with updating values over time, information on proteinuria was available for 144 patients, on statin use for 155, on ACE inhibitor use for 156, on smoking habit for 117, and on BMI for 143. Neither updating nor baseline values of all except proteinuria were associated with any significant effect on outcomes. Patients with baseline proteinuria >1 g/d had a four-fold higher risk for reaching ESRD (HR 4; 95% CI 1.17 to 14). However, results did not change when proteinuria was accounted for in the regression model of time to ESRD.
PKD1 occurs more frequently than PKD2 (
). The identification of PKD1 and PKD2 in the same family (P10), although a novel observation, would be expected in 1:250,000 to 1000,000 marriages in the general population ( 25 ). The observation that three PKD2 families from the same region had the same disease-associated haplotype but two different mutations was unexpected. This suggests that the disease-associated haplotype occurred frequently in the founder population and that two separate mutations occurred, each linked to the same haplotype. In Bardet-Biedl syndrome, we have made a similar observation in that the BBS1 mutation M390R arose in a haplotype that occurred frequently in the founder population and that both the same ancestral wild-type and disease-associated haplotype were introduced to Newfoundland in founder groups ( 24 ). 26
Figure 2 summarizes the important prognostic data identified in this longitudinal study of PKD1 and PKD2 cases. For screening and genetic counseling purposes, the most useful information is the youngest age at which the clinical event of interest occurred and the median age to the event. Although the incidence of hypertension that required treatment was frequent in both PKD1 and PKD2, ESRD occurred less frequently and at a later age in PKD2 compared with PKD1. Nonetheless, life expectancy in PKD2 was compromised, an observation also made by Hateboer et al. ( ). In the PKD1 group, life expectancy was extended by renal replacement therapy, and median life expectancy was almost similar to that in PKD2 (67 3 versus 71 yr). This observation is consistent with the fact that a lower mortality rate has been found in patients with ADPKD and ESRD compared with nondiabetic control patients with ESRD ( ). The observation that ESRD occurred earlier in PKD1 compared with PKD2 is not novel ( 27 2 , ), but the lifetime probabilities of developing hypertension that requires treatment and of development of stage 3 CKD are new and of clinical importance. 3
Several potential predictors of adverse outcome were assessed. In PKD1, variability in time to hypertension treatment and ESRD was observed between families. This was not attributable to a family history of essential hypertension. It suggests that different mutations or modifier genes influence the progression of renal disease in ADPKD. In PKD1, the position of the mutation correlates with the severity of renal disease (
), and in PKD2, patients with splice-site mutations seem to have milder renal disease compared with patients with other mutation types ( 28 ). 8
A gender effect on renal survival that favors female patients has been reported in PKD2 but not in PKD1 (
), and we have suggested that the gender of the parent who transmits PKD might be a risk factor ( 3 ). However, this study did not identify gender, gender of that parent who transmitted PKD, parity, or use of the OCP as risk factors for renal events in PKD1 or PKD2. Some of these conclusions are limited by the relatively small numbers of cases studied, particularly of women with PKD1 and PKD2. 10
This study has others limitations besides sample size. The incidence rates of stage 3 CKD are influenced by the fact that, at first serum creatinine measurement, CKD was already prevalent in 20% of PKD1 and in 38% of PKD2 (
Figure 1). It is likely that median age to onset of stage 3 CKD is earlier than the incidence rates observed in our groups, particularly for PKD2. Nine individuals with PKD1 and 16 with PKD2 had estimated GFR between 30 and 60 ml/min at first measurement and were excluded from the assessment of age to onset of CKD. A third limitation is that the 18 probands, although representative of the population, were subject to referral bias, because they were identified through the province’s nephrology/urology clinics.
Another limitation concerns study design. Family members were enrolled at different ages and phases of their disease. Accurate data on age to onset of each renal event was feasible, and potential risk factors that were present before the event happened were analyzed. Potential confounding factors such as proteinuria, smoking, BMI, and therapy with ACE inhibitors or statins were assessed and did not alter the conclusions. However, the presence of proteinuria of >1 g/d was an adverse risk factor for ESRD, which is consistent with current knowledge (
). Despite these limitations, the study has several advantages. It is population based, little ascertainment bias has occurred in family members studied, families were large, follow-up was prospective and long (22 yr), and relative risks that were observed for most risk factors studied were relatively small. 31 Conclusion
Hypertension that required treatment was frequent and occurred at an early age in both PKD1 and PKD2. CKD occurred later, and ESRD was infrequent in PKD2. Longevity was enhanced substantially by dialysis and transplantation in PKD1, such that life expectancy was almost as long as that in PKD2. Besides genotype, variability in outcomes was observed between families, and proteinuria was an adverse risk factor for ESRD. However, gender, gender of parent who transmitted PKD1, family history of essential hypertension, multiparity, and use of the OCP were not identified as risk factors for renal events in PKD.
Estimated GFR by age at first serum creatinine in family members with PKD1 (A) and PKD2 (B).
Summary of age to onset of renal events in PKD1 (A) and PKD2 (B). HT, hypertension; CKD, stage 3 chronic kidney disease.
Genetic and clinical characteristics of families with PKD1 and PKD2 a
Age to hypertension treatment in family members with PKD1
versus PKD2 and in unaffected family members a Table 3:
Risk factors for hypertension treatment in PKD a
Median age to onset of hypertension treatment by family in PKD1
Age to onset of CKD stage 3 and ESRD in PKD1
versus PKD2 Table 6:
Risk factors for ESRD in PKD1
Age to death in PKD1
versus PKD2 in cyst-positive versus cyst-negative family members
P.R. holds a young investigator award from the Italian Society of Nephrology; W.S.D. and Y.P. received funding from Canadian Institutes of Health Research (CIHR) and Kidney Foundation of Canada; P.S.P. holds a CIHR Regional Partnership Program Distinguished Scientist Award and received funding from the Janeway Hospital Foundation.
We thank the patients and families who made this study possible. E.D. acknowledges the Health Care Corporation of St. John’s for excellent support during this study.
Published online ahead of print. Publication date available at
. www.jasn.org References
1. Nauli SM, Alenghat FJ, Luo Y, Williams E, Vassilev P, Li X, Elia AE, Lu W, Brown EM, Quinn SJ, Ingber DE, Zhou J: Polycystins 1 and 2 mediate mechanosensation in the primary cilium of kidney cells. Nat Genet 33: 129–137, 2003
2. Parfrey PS, Bear JC, Morgan J, Cramer BC, McManamon PJ, Gault MH, Churchill DN, Singh M, Hewitt R, Somlo S, Reeders ST: The diagnosis and prognosis of autosomal dominant polycystic kidney disease. N Engl J Med 323: 1085–1090, 1990
3. Hateboer N, v Dijk MA, Bogdanova N, Coto E, Saggar-Malik AK, San Millan JL, Torra R, Breuning M, Ravine D: Comparison of phenotypes of polycystic kidney disease types 1 and 2. Lancet 353: 103–107, 1999
4. Milutinovic J, Fialkow PJ, Agodoa LY, Phillips LA, Rudd TG, Bryant JI: Autosomal dominant polycystic kidney disease: Symptoms and clinical findings. QJ Med 53: 511–522, 1984
5. Hateboer N, Lazarus PL, Williams AJ, Holmans P, Ravine D: Familial phenotype differences in PKD1. Kidney Int 56: 34–40, 1999
6. Kelleher CL, McFann KK, Johnson AM, Schrier RW: Characteristics of hypertension in young adults with autosomal dominant polycystic kidney disease compared with the general US population. Am J Hypertens 17: 1029–1034, 2004
7. Gabow PA, Johnson AM, Kaehny WD, Kimberling WJ, Lezotte DC, Duley IT, Jones RH: Factors affecting the progression of renal disease in autosomal dominant polycystic kidney disease. Kidney Int 41: 1311–1319, 1992
8. Magistroni R, He N, Wang K, Andrew R, Johnson A, Gabow P, Dicks E, Parfrey P, Torra R, San-Millan JL, Coro E, Van Dirk M, Breuning M, Peters D, Bogdanova N, Ligabue G, Albertazzi A, Hateboer N, Demetriou K, Pierides A, Deltas C, St. George-Hyslop P, Ravine D, Pei Y: Genotype-renal function correlation in type 2 autosomal dominant polycystic kidney disease. J Am Soc Nephrol 14: 1164–1174, 2003
9. Gretz N, Zeier M, Geberth S, Strauch M, Ritz E: Is gender a determinant for evolution of renal failure? A study in autosomal dominant polycystic kidney disease. Am J Kidney Dis 14: 178–183, 1989
10. Bear JC, Parfrey PS, Morgan JM, Martin CJ, Cramer BC: Autosomal dominant polycystic kidney disease: New information for genetic counseling. Am J Med Genet 43: 548–553, 1992
11. Stewart JH: End-stage renal failure appears earlier in men than in women with polycystic kidney disease. Am J Kidney Dis 24: 181–183, 1994
12. Simon P, Le Goff JY, Ang KS, Charasse E, Le Cacheus P, Cam G: Epidemiologic data, clinical and prognostic features of autosomal dominant polycystic kidney disease in a French region. Nephrologie 17: 123–130, 1996
13. Schrier RW, Johnson AM, McFann K, Chapman AB: The role of parental hypertension in the frequency and age of diagnosis of hypertension in offspring with autosomal-dominant polycystic kidney disease. Kidney Int 64: 1792–1799, 2003
14. Geberth S, Stier E, Zeier M, Mayer G, Rambausek M, Ritz E: More adverse renal prognosis of autosomal dominant polycystic kidney disease in families with primary hypertension. J Am Soc Nephrol 6: 1643–1648, 1995
15. Jones DC, Hayslett JP: Outcome of pregnancy in women with moderate or severe renal insufficiency. N Engl J Med 335: 226–232, 1996
16. Kang AK, Duncan JA, Cattran DC, Floras JS, Lai V, Scholey JW, Miller JA: Effect of oral contraceptives on the renin angiotensin system and renal function. Am J Physical Regul Integr Comp Physiol 280: R807–R813, 2001
17. Parfrey PS, Davidson WS, Green JS: Clinical and genetic epidemiology of inherited renal disease in Newfoundland. Kidney Int 61: 1925–1934, 2002
18. Barrett BJ, Foley R, Morgan J, Hefferton D, Parfrey P: Differences in hormonal and renal vascular responses between normotensive patients with autosomal dominant polycystic kidney disease and unaffected family members. Kidney Int 46: 1118–1123, 1994
19. Pei Y, Watnick T, He N, Wang K, Laing Y, Parfrey P, Germino G, St. George-Hyslop P: Somatic PKD2 mutations in individual kidney an liver cysts supports a “two-hit” model of cystogenesis in type 2 autosomal dominant polycystic kidney disease. J Am Soc Nephrol 10: 1524–1529, 1999
20. Watnick T, He N, Wang K, Laing Y, Parfrey P, Hefferton D, St. George-Hyslop P, Germino G, Pei Y: Mutations of PKD1 in ADPKD2 cysts suggest a pathogenic effect of trans-heterozygous mutations. Nat Genet 25: 143–144, 2000
21. O’Dea D, Murphy SW, Hefferton D, Parfrey PS: Higher risk for renal failure in first-degree relatives of white patients with end stage renal disease: A population-based study. Am J Kidney Dis 32: 794–801, 1998
22. Ravine D, Gibson RN, Walker RG, Sheffield LJ, Kincaid-Smith P, Danks DM: Evaluation of ultrasonographic diagnostic criteria for autosomal dominant polycystic kidney disease. Lancet 343: 824–826, 1994
23. Levey S, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D: A more accurate method to estimate glomerular filtration rate from serum creatinine: A new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 130: 461–470, 1999
24. Pei Y, Paterson AD, Wang KR, He N, Hefferton D, Watnick T, Germino G, Parfrey P, Somlo S, St. George-Hyslop P: Bilineal disease and trans-heterozygotes in autosomal dominant polycystic kidney disease. Am J Hum Genet 68: 355–363, 2001
25. Igarashi P, Somlo S: Genetics and pathogenesis of polycystic kidney disease. J Am Soc Nephrol 13: 2384–2398, 2002
26. Fan Y, Green JS, Ross AJ, Beales PL, Parfrey PS, Davidson WS: Linkage disequilibrium mapping in the Newfoundland population: A re-evaluation of the refinement of the Bardet-Biedl syndrome 1 critical interval. Hum Genet 116: 62–71, 2005
27. Perrone RD, Ruthzaer R, Terrin NC: Survival after end-stage renal disease in autosomal dominant polycystic kidney disease: Contribution of extrarenal complications to mortality. Am J Kidney Dis 38: 777–784, 2002
28. Rossetti S, Burton S, Strmecki L, Pond GR, San Millan JL, Zerres K, Barratt TM, Ozen S, Torres VE, Bergstralh EJ, Winearls CG, Harris PC: The position of the polycystic kidney disease 1 (PKD1) gene mutation correlates with the severity of renal disease. J Am Soc Nephrol 13: 1230–1237, 2002
29. Kaplan EL, Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 53: 457–481, 1958
30. Norman GR, Streiner DL: Biostatistics: The Bare Essentials, Toronto, Mosby, 1994, pp 236–254
31. Jafar TH, Stark PC, Schmid CH, Landa M, Maschio G, Marcantoni C, de Jong PE, de Zeeuw D, Shahinfar S, Ruggenenti P, Remuzzi G, Levey AS; AIPRD Study Group. Angiotensin-converting enzymne inhibition and progression of renal disease: Proteinuria as a modifiable risk factor for the progression of non-diabetic renal disease. Kidney Int 60: 1131–1140, 2001