To analyze the effect of treatment with either pulse cyclophosphamide (CYC) or azathioprine (AZA) combined with methylprednisolone (MP), on serial biopsy results in patients with proliferative lupus nephritis, and to evaluate the predictive value of various histopathologic and clinical parameters with regard to disease outcome.
Biopsy specimens from patients with proliferative lupus nephritis, obtained at study entry and after 2 years of therapy, were scored according to a standardized method, and results assessed in relation to disease outcome.
Of the 87 patients originally enrolled, 39 underwent repeat biopsy. These patients were representative of the overall group, both at entry and at 2-year followup. The median activity index changed from 8.0 to 2.7 (no differences between the treatment groups). In the group treated with AZA plus MP (AZA group), the increase in the median chronicity index (from 2.7 to 3.8) was larger than that in the CYC group (from 2.7 to 3.0) (P = 0.050). In multivariate analyses, renal function at enrollment and after 2 years was the best predictor of renal function at the last visit, while none of the histopathologic variables (either at entry or at 2 years) added to this prediction. Comparing patients whose disease transitioned to class II with those who had persistent proliferative lupus nephritis revealed no differences between the treatment groups at either time point, and no clinical differences at 2 years. However, a higher serum creatinine level at entry and greater proteinuria at last visit were characteristic of patients who still had proliferative lupus nephritis at 2 years.
These results indicate that, although both CYC and AZA are effective in reducing active lesions in lupus nephritis, progression of chronic lesions is more effectively halted by CYC. Variables assessed by repeat biopsy do not predict clinical outcome.
The prognosis of patients with proliferative lupus nephritis has improved considerably since the institution of treatment regimens consisting of corticosteroids combined with cytotoxic immunosuppressive drugs (1). However, many side effects of this intensive treatment also occur (1). Individualized, patient-specific treatment plans are therefore needed.
To identify patients who are at risk of developing end-stage renal disease, clinical, biochemical, and histopathologic predictors have been sought. Clinical and biochemical parameters identified in previous studies have included age (2), sex (2, 3), race (4–6), serum creatinine level at presentation (4, 7–11), anti-SSA antibodies (11), and the Systemic Lupus Erythematosus Disease Activity Index (9, 12). In many studies, the predominant renal biopsy findings that predicted worse outcome were chronic lesions or the total chronicity index (CI) (2, 4, 5, 8, 10, 13–15). The activity index (AI) predicted outcome in some studies (4, 5, 7, 13, 16). However, other studies did not show either of the indices to be of prognostic value (6, 17–21).
In several reported studies, repeat or serial renal biopsies of patients with lupus nephritis were performed. In most of these studies, repeat biopsies were obtained for clinical reasons, i.e., when clinical renal parameters worsened (3, 15, 22–28). The transition in World Health Organization (WHO) lupus nephritis class has been the focus of several studies (23–25, 29–31). Others have evaluated the effects of treatment and the change in the AI and the CI (20, 32–38), or whether the treatment regimen was modified based on findings in the repeat biopsy (39). It has been suggested that repeat biopsies are of more prognostic importance than initial biopsies (21, 40).
The first Dutch Lupus Nephritis Study was initiated with the goal of seeking a less toxic alternative to pulse cyclophosphamide (CYC) therapy. Patients were enrolled beginning in 1995. Eighty-seven systemic lupus erythematosus (SLE) patients with biopsy-proven proliferative lupus nephritis were randomized to receive either pulse CYC therapy or azathioprine (AZA) combined with methylprednisolone (MP) pulses (41). A repeat renal biopsy was performed in 39 participants, according to a predetermined timetable (i.e., after 2 years of treatment). For the present pathology study, a standardized scoring form was used to evaluate the changes in renal biopsy parameters and WHO class. The aim was to compare biopsy parameters in these patients at study entry and after 2 years, and to analyze the differences between the 2 treatment arms. Moreover, the predictive value of laboratory and histopathologic parameters at study entry and after 2 years of treatment was studied.
PATIENTS AND METHODS
Patients in the first Dutch Lupus Nephritis Study were enrolled between September 1995 and September 2001. All patients in this randomized controlled trial (n = 87) had biopsy-proven proliferative lupus nephritis. Patients were randomly assigned to receive 2-year treatment with either pulse CYC (13 pulses) plus oral prednisone (CYC group) or AZA plus pulse MP (9 pulses) and oral prednisone (AZA group). After 2 years the treatment regimen was switched to long-term AZA plus prednisone in all patients. All patients were followed up regularly, and data on clinical and laboratory parameters were recorded according to the study protocol (41). Laboratory measures included serum creatinine level, estimated glomerular filtration rate (GFR) (42), proteinuria, C3 and C4 complement levels, and anti–double-stranded DNA (anti-dsDNA) titer at the time of the biopsy. Outcome measures were the occurrence of complete remission, doubling of the serum creatinine level, treatment failure or relapse, and levels of serum creatinine and proteinuria at the last visit.
The first biopsy in each patient was performed before randomization for study treatment. All 87 biopsy samples were classified according to the WHO 1995 criteria for lupus nephritis (43). After 2 years, all available patients were asked to undergo a repeat renal biopsy for study purposes only, the results of which would be made anonymous and not used for treatment decisions. The ethics committees of all participating hospitals approved this pathology study. After having given their informed consent, 41 patients underwent a repeat biopsy. Two of these repeat biopsies were excluded since they did not contain any glomeruli. Thus, for this study 126 biopsy specimens were used.
All biopsies were performed and processed locally. For this study, they were scored for several glomerular, tubulointerstitial, and vascular features (listed in Appendix A), by 5 nephropathologists. Both the AI and the CI for lupus nephritis were estimated (13), and both the initial and the repeat biopsy samples were classified according to the WHO 1995 classification.
For definitions of the histopathologic features, see Appendix A. The following clinical definitions were used: primary study end point = (not necessarily sustained) doubling of the serum creatinine level at any time point during followup; primary treatment failure = doubling of baseline serum creatinine level, evaluated at week 12; relapse = doubling (from week 12 onward) of the lowest serum creatinine value obtained to date and/or development of either nephrotic syndrome (proteinuria >3.5 gm/day and serum albumin <30 gm/liter) with the lowest protein excretion value obtained to date ≤2.0 gm/day repeatedly, or proteinuria >1.5 gm/day without other causes, in a previously nonproteinuric patient; complete remission = serum creatinine <130% of the lowest serum creatinine level since the start of treatment, proteinuria <0.5 gm/day, and <10 erythrocytes per high-power field.
All data were analyzed using SPSS 12.0.1 software. All patients were followed up in an intent-to-treat analysis. Descriptive statistics included frequency tables of the histopathologic and clinical characteristics. Glomerular parameters were expressed as the percentage of the total number of nonsclerotic and nonischemic glomeruli present on the slide. Comparisons and correlations were studied with nonparametric tests.
Intraclass correlation coefficients (ICCs) were calculated (44), using all available biopsy samples. In general, values >0.8 were considered to indicate excellent concordance, values of 0.6–0.8 good concordance, values of 0.4–0.6 moderate concordance, and values <0.4 poor concordance. For the analyses of the changes between biopsies obtained at study entry and those obtained after 2 years of therapy and for the analyses of the prognostic value of histopathologic items, only those parameters that were frequently present and had an ICC >0.4 were used. These are summarized in Appendix B. Each biopsy specimen was scored by 3 of 5 nephropathologists, and the mean value of the scores was used for the analyses.
Differences between the 2 treatment groups with regard to histopathologic changes between the first and the second biopsy were studied with repeated-measurements mixed model tests. The occurrence of the events doubling of serum creatinine level, treatment failure or relapse, and achievement of complete remission was analyzed by stepwise logistic regression and expressed as Nagelkerke R2 (45). The time to occurrence of these events was analyzed using Cox regression analysis. The outcome variables serum creatinine level at last visit and proteinuria at last visit were analyzed using stepwise multivariate regression methods. Since there was not a normal distribution, these parameters were transformed (log- or square root–) before being entered into the equation. Predictive models are expressed as adjusted R2. The last observation carried forward method was used, except in the case of patients who were lost to followup during the first 2 years (n = 2). All other patients, including those who experienced a renal relapse, were considered in the intent-to-treat analysis.
For the independent histopathologic and clinical variables, the P value to enter the diverse regression analyses was set at <0.05, and the P value to remain was set at <0.10. To study the possible effect of the 2 treatment regimens, treatment was added to all analyses, as an independent parameter.
P values less than or equal to 0.05 were considered significant. To account for the effect of multiple comparisons, the Bonferroni correction was applied to correlation coefficients; P values less than 0.01 were considered significant. Uncorrected P values are reported in the tables.
At the beginning of 2006, the patients in the study had been followed up for a median of 77 months (interquartile range [IQR] 54–96). The primary study end point (doubling of the serum creatinine level [not necessarily sustained]) was reached in 9 patients; this occurred more often in the AZA group than in the CYC group (relative risk 5.2, 95% confidence interval 1.1–25.2, P = 0.040). Primary treatment failure was observed in 1 patient, who began hemodialysis at week 12 and died 9 years later due to intracerebral abscesses after renal transplantation, without showing any sign of lupus activity. Renal relapses occurred in 17 patients. Eight of these relapses occurred during the first 2 years of treatment. Renal relapses occurred significantly more frequently in the AZA-treated patients than in the CYC-treated patients (relative risk 4.9, 95% confidence interval 1.6–15.0, P = 0.006). Seven patients died, and 5 were lost to followup. The median serum creatinine level at last followup was 82 μmoles/liter (IQR 74–108), and median proteinuria at last followup was 0.30 gm/24 hours (IQR 0.10–0.72) (no significant difference between the 2 treatment arms).
At last visit most patients (65%) were still taking low doses of AZA and prednisone. Other immunosuppressive regimens were AZA alone (4%), prednisone alone (20%), or mycophenolate mofetil combined with prednisone (8%), while 4% of the patients had discontinued all immunosuppressive drugs. At the time of last visit 22% of the patients were receiving statins. The median systolic and diastolic blood pressure at last visit were 120 mm Hg (IQR 110–130) and 80 mm Hg (IQR 70–83), respectively. The majority of patients (70%) were taking angiotensin-converting enzyme (ACE) inhibitors or angiotensin receptor blockers, and 39% used other or additional antihypertensive drugs. The mean number of antihypertensive drugs (including ACE inhibitors and angiotensin receptor blockers) was 1.3 (SD 1.1). There were no differences in treatment regimen at last visit between the CYC and AZA groups.
Histopathologic characteristics at study entry.
The histopathologic characteristics at study entry are summarized in Table 1. The median percent of glomeruli with extracapillary proliferation was 13%. This was more often segmental than circumferential (85% versus 15%). Fibrous crescents were rarely seen. The median activity and chronicity indexes were 9.3 and 2.7, respectively. There were no differences in histologic findings between the 2 treatment groups, except that endocapillary proliferation (especially global) and interstitial fibrosis were more frequent in the CYC group (P = 0.012 and P = 0.008, respectively) (data not shown).
Table 1. Histopathologic and clinical characteristics of all lupus nephritis patients included in the study and of the patients who underwent a repeat biopsy after 2 years of treatment*
All patients (n = 87)
Patients with repeat biopsy (n = 39)
Only items with moderate to good interobserver agreement (intraclass correlation coefficient >0.4) were used for the analyses. Except where indicated otherwise, values are the percent of patients. IQR = interquartile range; WHO = World Health Organization; CYC = cyclophosphamide; AZA = azathioprine; anti-dsDNA = anti–double-stranded DNA.
Fibrous crescents are scored in the chronicity index, while cellular extracapillary proliferation is a component of the activity index.
Spikes were scored only if they covered >50% of the glomerular area.
P = 0.02 versus patients who did not undergo a repeat biopsy.
Not all biopsy specimens contained enough glomeruli to be classified, and 2 were not classified as showing proliferative lupus nephritis.
To convert creatinine values to milligrams per deciliter, divide by 88.4.
Because 9 patients were no longer available after 2 years of treatment, n = 78 for the group “all patients.”
Although the patients who presented with lupus nephritis as their first symptom of SLE (n = 36) had fewer chronic changes evidenced on the biopsy, their serum creatinine levels were higher than those in patients who had been diagnosed as having SLE previously (P = 0.03). Also, karyorrhexis was more prevalent in patients newly diagnosed as having SLE (P = 0.004).
Histopathologic abnormalities, which were more often seen in patients with worse renal function at presentation (GFR <70 ml/minute), consisted mainly of tubulointerstitial lesions: interstitial infiltration, interstitial fibrosis, and tubular atrophy. C3 levels were negatively associated with the presence of some types of active lesions (karyorrhexis r = −0.389, P < 0.01; endocapillary proliferation r = −0.317, P < 0.01), and positively associated with the total CI (r = 0.340, P < 0.01). The presence of anti-dsDNA antibodies was positively correlated with the total AI (r = 0.351, P < 0.01), extracapillary proliferation (r = 0.443, P < 0.01), and loop necrosis (r = 0.354, P < 0.01), but negatively associated with karyorrhexis (r = −0.389, P < 0.01).
Compared with patients with a low CI (<4; n = 73), patients with a high CI (≥4; n = 14) more frequently had hypertension (P = 0.018). Worse renal function was also seen in patients with a high CI (P = 0.045), while no significant differences were found with regard to proteinuria. Apart from the separate components of the CI, patients with a high CI also more frequently had synechia (P = 0.001), another parameter of chronic damage. Patients with a high AI (≥10; n = 38) more frequently had an active urine sediment (P = 0.012) and higher anti-dsDNA titers (P = 0.01) than patients with a low AI (<10; n = 49). There were no significant differences between the high AI and low AI groups with regard to proteinuria or renal function. Besides the separate components of the AI, more loop necrosis was present in biopsy specimens from patients with a high AI (P = 0.000). The percentage of sclerosed glomeruli was also lower in samples from patients with a high AI, but this difference did not quite reach statistical significance (P = 0.054).
Characteristics of the patients who underwent repeat biopsy.
Characteristics of the patients who underwent repeat biopsy are shown in Table 1. Both at study entry and after 2 years of therapy, there were no differences in clinical or laboratory characteristics between the subgroup of patients who underwent repeat biopsy and those who did not. Apart from a significantly, but only slightly, lower percentage of ischemic glomeruli in specimens from the group of patients who underwent a repeat biopsy, there were no differences in baseline histopathologic characteristics. Moreover, the frequency of primary treatment failure or renal relapse during the first 2 years of treatment did not differ between the group of patients who did and those who did not undergo repeat biopsy.
Histopathologic changes during treatment.
At repeat renal biopsy (n = 39), a decrease in the number of active lesions was seen, with an increase in chronic damage (Table 2): the median AI decreased from 8.0 to 2.7 (P < 0.001), and the CI increased from 2.7 to 3.3 (P < 0.001). Figure 1 depicts the changes in both the AI and the CI, overall and by treatment group. The effects of the 2 treatment regimens were significantly different with regard to the extent of completely sclerosed glomeruli, tubular atrophy, and the CI (P = 0.042, P = 0.007, and P = 0.050, respectively). These differences were all in favor of the CYC regimen. The median CI increased from 2.7 to 3.0 in the CYC group, while it increased from 2.7 to 3.8 in the AZA group. The percentage of completely sclerosed glomeruli increased from 2 to 9 in the CYC group and from 7 to 14 in the AZA group. The percentage of repeat biopsy specimens showing tubular atrophy covering ≥25% of the slide was 16% and 40% in the CYC group and the AZA group, respectively. The extent of the decrease in the AI or its components was not different between the 2 treatment arms.
Table 2. Change in histopathologic characteristics after 2 years of therapy, in 39 patients*
After 2 years
Only items with moderate to good interobserver agreement were used for the analyses. Except where indicated otherwise, values are the median (interquartile range). NS = not significant.
Component of the activity index.
Fibrous crescents are scored in the chronicity index, while cellular extracapillary proliferation is a component of the activity index. Fibrous crescents were rare; therefore, extracapillary proliferation is included under active lesions.
Component of the chronicity index.
Change from the first to the second biopsy was significantly different between the 2 treatment groups.
Of the 39 repeat biopsies, 5 did not include enough glomeruli to be classified and 2 could not be classified because of poor quality of the slides, leaving 32 biopsy samples for evaluation. In most biopsies a transition was seen: 12 (38%) changed from proliferative lupus nephritis (WHO class III or IV) to class II, and 13 (41%) changed from class IV to class III. The transitions are summarized in Figure 2. Although more CYC-treated patients had a transition to WHO class II (n = 8) as compared with the AZA-treated patients (n = 4) and more patients in the AZA group still had proliferative lupus nephritis after 2 years (12 patients versus 7 patients), the distribution of these changes between the treatment arms was not significantly different (P = 0.149).
With the International Society of Nephrology/Renal Pathology Society 2003 classification for lupus nephritis (46), an equal distribution of the segmental and global nephritis subtypes was also found in patients from the 2 treatment arms, both in the first and second biopsies. In patients with persistent proliferative lesions on repeat biopsy (n = 19) compared with those who transitioned to class II (n = 12), we found no differences in clinical characteristics at study entry. Although patients with persistent proliferative lupus nephritis had higher serum creatinine levels at study entry, their serum creatinine levels both at repeat biopsy and at last visit were not different from those in patients with histologic improvement. The patients with persistent proliferative lesions had higher proteinuria at last visit (0.16 gm/24 hours versus 0.05 gm/24 hours), but not at the other time points. Some of the laboratory and histologic characteristics are summarized in Table 3.
Table 3. Comparison of patients with persistent proliferative lesions at repeat biopsy versus those with a transition to nonproliferative lesions (WHO class II)*
Proliferative lesions on repeat biopsy (n = 19)
Nonproliferative lesions on repeat biopsy (n = 12)
One patient had a transition to World Health Organization (WHO) class VI and was excluded from this analysis; the remaining 7 biopsies could not be classified because they did not contain enough glomeruli (n = 5) or because of poor quality of the slides (n = 2). Except where indicated otherwise, values are the median (interquartile range). NS = not significant; anti-dsDNA = anti–double-stranded DNA.
To convert creatinine values to milligrams per deciliter, divide by 88.4.
Association between clinical or biopsy parameters and outcome.
Doubling of serum creatinine level and treatment failure or relapse.
Neither doubling of the serum creatinine level nor treatment failure or relapse was predicted by clinical or biopsy parameters at study entry. Treatment with AZA plus MP, however, was strongly associated with these events (Nagelkerke R2 0.207 [P = 0.025] for doubling of the serum creatinine level at any time point during followup, and 0.226 [P = 0.003] for treatment failure or relapse). Also for time to doubling of the serum creatinine level, treatment failure, or relapse, treatment was the only predictive factor. At the time of repeat biopsy, however, none of the 2-year parameters, including treatment, could predict the occurrence of doubling of the serum creatinine level, treatment failure, relapse, or time to these events.
Per the study protocol, the 2 groups underwent different treatment regimens during the first 2 years. To evaluate a difference in the efficacy of the initial treatment, remission was assessed for this period only. Thus, for these analyses, only the first biopsy was used. Attainment of complete remission was associated with the percentage of completely sclerosed glomeruli (Nagelkerke R2 0.140 [P = 0.029]). Of the clinical parameters, only proteinuria was associated with reaching complete remission (an association that would be expected, since proteinuria [<0.5 gm/24 hours] was one of the criteria for complete remission). The time to complete remission was also associated with the percentage of sclerosed glomeruli. Remission was not predicted by treatment with AZA or CYC.
Parameters at last visit.
Of the clinical and histopathologic parameters assessed at study entry, the combination of renal function, percentage of completely sclerosed glomeruli, and degree of extracapillary proliferation was predictive of the serum creatinine level at last visit (adjusted R2 for these 3 parameters 0.248). The serum creatinine level at the time of repeat biopsy predicted the serum creatinine level at last followup. The presence of tubular casts on repeat biopsy was also a significant predictor of the serum creatinine level at the last visit, although the association was not as strong (adjusted R2 for these 2 parameters 0.410). Neither clinical parameters nor histopathologic parameters at the time of repeat biopsy were predictive of proteinuria at the last visit. The percentage of completely sclerosed glomeruli and the presence of tubular casts in the first biopsy sample, however, were significant predictors of the degree of proteinuria at last visit (adjusted R2 for the 2 parameters 0.216). The type of treatment was not a predictor of the outcome parameters serum creatinine level or proteinuria at last visit.
When the parameters renal function at study entry, renal function at repeat biopsy, treatment, percentage of sclerosed glomeruli in the first biopsy sample, and crescents in the first biopsy sample were entered into a predictive model, it was found that the serum creatinine level at last visit was best predicted by the combination of serum creatinine level at 2 years and the percentage of sclerosed glomeruli in the first biopsy sample (adjusted R2 for this combination 0.585; adjusted R2 for serum creatinine level at 2 years alone 0.546). Treatment did not add predictive value in this equation.
In our representative cohort of patients with proliferative lupus nephritis who underwent repeat biopsy after 2 years of induction therapy, a significant decrease in the AI and a significant increase in the CI were observed, both in patients treated with CYC and in those treated with AZA plus MP. The increase in chronic changes was, however, significantly greater in the AZA group. This indicates that treatment for 24 months with 13 pulses of CYC is more effective in tempering the progression of chronic lesions than is 24-month treatment with AZA and 9 pulses of MP. If the CYC group had not had greater interstitial fibrosis at study entry, the observed significant differences with regard to the CI would probably have been larger, emphasizing the superior efficacy of CYC. This is also reflected by the lower incidence of relapse and of doubling of the serum creatinine level in the CYC group. Whether AZA plus MP is also less effective compared with a short course of CYC, low-dose CYC, or mycophenolate mofetil remains to be investigated; whether pulse CYC for 24 months is more effective than treatment with mycophenolate mofetil is another relevant question.
Although the present results indicate that CYC administered for 2 years is more effective with regard to both renal relapses and chronic lesions, the evidence obtained to date suggests that long-term renal function in patients receiving AZA plus MP is comparable with that in patients treated with CYC. Therefore, for selected patients with proliferative lupus nephritis who wish to avoid the gonadal toxicity of CYC, treatment with AZA plus MP can be an alternative.
Figure 3 shows the degree of change in the CI between first biopsy and repeat biopsy, in the present study and in previous studies of adult lupus nephritis patients in whom repeat biopsies were performed according to a predetermined timetable (20, 32–37). Since the CI was introduced in 1984, studies published before that date were not considered. It should be noted that comparison of the studies is hampered by a number of factors: not all studied patients were diagnosed as having proliferative lupus nephritis, different treatment regimens were used, and the publications varied in terms of reporting either means or medians. A difference in ethnicity could also play a role in the observed differences. Not all studies of lupus nephritis patients who underwent repeat biopsies according to standard protocols could be included in the Figure 3, since the data in some reports were incomplete. Performing a genuine meta-analysis on these data was therefore impossible. Nevertheless, the figure provides an impression of the observed change in the CI during or after induction treatment. As can be seen, the change in CI in our study population was quite small compared with that in other studies.
In 2 previous studies, it was found that persistently reduced complement levels were associated with a larger increase in the CI and/or worse renal outcome (14, 38). In the present study, however, neither depression of the C3 level nor the duration of depressed C3 levels was correlated with the (change in) CI.
In our study population, the AI was >3 in 14 repeat biopsies (36%), and none of the biopsies showed inactive disease (AI 0). Of the patients in whom a repeat biopsy was performed, 1 fulfilled the clinical criteria for renal relapse at the time of biopsy and was therefore treated with reinduction therapy, as was prescribed by the study protocol. The AI and CI on second biopsy in this particular patient were 7.7 and 5.7, respectively. All other patients had clinically silent disease.
This discrepancy between clinically quiescent disease and the presence of active lesions on biopsy has been observed by others. Favre et al found a mean ± SD AI of 3.6 ± 0.5 in 26 repeat biopsies of patients with lupus nephritis (WHO class varying from II to V) after 18–24 months of treatment with cyclosporin A and prednisone (47). Nevertheless, in general a decline in the AI has been observed on repeat biopsy after treatment; Pillemer et al noted improvement in the AI in 71% of their patients after a minimum of 2 years of treatment (38), and patients studied by Esdaile and colleagues had a median AI of 2.0 in biopsies obtained after 2 years of treatment (33). Many studies have addressed the transition in WHO class in lupus nephritis patients who have undergone repeat biopsies either due to clinical deterioration (3, 22, 27, 39) or as part of a predetermined study protocol to assess the results of treatment (34, 35, 48). In our patients in whom a repeat biopsy was performed, a transition to less active disease was generally seen on the repeat biopsy, although 59% were still classified as having proliferative lupus nephritis, predominantly class III. This is in accordance with most other reports.
The percentage of sclerotic glomeruli in the biopsy at presentation was the best prognostic factor for predicting the course of the disease, i.e., occurrence of complete remission, time to attainment of complete remission, and serum creatinine level and proteinuria at last visit. This is consistent with previous studies (49). Notably, there was no correlation between the percentage of completely sclerosed glomeruli and renal function at the time of biopsy (r = −0.182 for GFR and r = 0.175 for serum creatinine level). This could best be explained by hyperfiltration of the remaining noninflamed glomeruli. Also of note, neither the AI nor the CI predicted renal outcome.
As has been observed by many others (4, 5, 7, 50), we found that renal function was predictive of outcome. Moreover, renal function after immunosuppressive treatment for proliferative lupus nephritis has previously been shown to be prognostic of long-term renal function (37, 51).
In contrast to our data, several authors have reported that findings on repeat biopsy were predictive of outcome (21, 33, 52). In 1993 Esdailet et al showed that subepithelial and subendothelial deposits on repeat biopsy were associated with both renal outcome and SLE-related mortality. Findings on light microscopy, however, were not significantly associated with outcome, although there was a trend toward an association of both the AI and the tubulointerstitial index with renal outcome (33). Hill et al studied repeat biopsies in 71 patients after 6 months of intravenous CYC therapy; in 18 of these patients, the serum creatinine level had doubled after a mean followup of 7 years (21). They showed that patients who had reversible interstitial fibrosis and segmental glomerular scarring had a better prognosis (52). The main differences between that study population and ours were that fewer of our patients were non-Caucasian (24% versus 44%) and that our patients received long-term maintenance treatment. Data on treatment after 6 months were not mentioned in the report by Hill and colleagues. In other studies findings on repeat biopsy were also predictive of renal outcome, but these biopsies had been obtained for clinical reasons, thereby introducing a selection bias (3, 15).
Interestingly, although the increase in the CI was larger in the AZA group than in the CYC group in the present study, the CI itself, either at study entry or on repeat biopsy, did not predict outcome. This could be explained by the overlap of CI values in the 2 treatment groups, or by the followup time of 77 months, which might be too short to identify differences in renal outcome.
In conclusion, our study showed that treatment of lupus nephritis with pulse cyclophosphamide for 24 months delays the increase in chronicity better than does treatment with azathioprine, and better than has been found in most previously reported studies. Whether newer therapies, such as mycophenolate mofetil, are better in retarding chronic lesions remains to be determined. The per-protocol repeat renal biopsy after 2 years of treatment, the results of which did not influence treatment decisions, did not provide additional information with regard to outcome at the last followup. Both renal function at entry and renal function after 2 years were better predictors of renal outcome than were the histopathologic features found on repeat biopsy. Thus, clinical parameters in patients with lupus nephritis are more informative than are findings on repeat renal biopsy. To compare the effects of treatment regimens repeat biopsies, performed according to a predetermined protocol, should be performed only in randomized clinical trials; our present results indicate that repeat biopsies in individual patients with proliferative lupus nephritis do not provide prognostic information.
Dr. Berden had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. Grootscholten, van Houwelingen, Derksen, Berden.
Statistical analysis. Grootscholten, van Houwelingen.
We would like to thank all patients and members of the Dutch Working Party on SLE who participated in the first Dutch Lupus Nephritis Study. This study was possible because of the skillful coordination by Dr. G. Ligtenberg (University Medical Center Utrecht). We thank the pathologists who supplied the renal biopsy specimens. We appreciate the help of Drs. J. J. Weening (Academic Medical Center), J. A. Bruijn (Leiden University Medical Center), and K. J. Assmann (Radboud University Nijmegen Medical Center), who participated in the initial development of the scoring form. Finally, we would like to thank Dr. G. S. Hill (Hôpital Europeen Georges Pompidou, Paris, France) for his input on the scoring form.
Table . DEFINITIONS OF HISTOLOGIC FEATURES
Scored using periodic acid–Schiff (PAS) staining.
Scored in both arteries and arterioles.
No. of glomeruli
Maximal number of glomeruli excluding those which are incomplete (e.g., at the edge)
No. of normal glomeruli
Completely sclerosed glomeruli
No. of completely sclerosed glomeruli
No. of ischemic but otherwise normal glomeruli (i.e., with wrinkling of all capillary basement membranes)
Increase in cells in the mesangium, with or without an increase in mesangial matrix
Increase in cellularity of the capillary walls; on top of mesangial proliferation; segmental or global (≤50% or >50% of Bowman's space)
≥3 cell layers between Bowman's capsule and capillary tuft; circumferential or segmental (>50% or ≤50% of Bowman's space); cellular (purely cellular), fibrocellular (cellular with the addition of fibrillar material), or fibrous (predominantly fibrotic [sclerotic] lesion with fibroblasts)
Any duplication of glomerular basement membrane, or so-called “tram tracking”
Extensions of glomerular basement membrane between deposits (egg racks); ≥50% of glomerulus
Disruption of the glomerular basement membrane, with or without presence of fibrin-rich material
Detachment of the glomerular basement membrane from the mesangial matrix, with formation of microaneurysms
Small fibrous adhesion of the flocculus with Bowman's capsule, without proliferation or sclerosis
Diffuse mesangial accumulation of silver-positive material; segmental or global (part or whole of glomerulus, within capillary tuft)
Macrophages in the tubular lumen (scored as present or absent)
Inflammatory cells in the interstitium, in unscarred areas (scored <5–10%, 10–25%, 25–50%, or >50%, with <5–10% indicating absent, or present but rare)
Inflammatory cells in the interstitium, in unscarred areas (scored <5–10%, 10–25%, 25–50%, or >50%, with <5–10% indicating absent, or present but rare)
Increased extracellular fluid in the interstitium resulting in increased spacing between tubules; a pale and reticulated appearance on trichrome staining of the interstitium (scored as present or absent)
Total occlusion with fibrin (scored as present or absent)
Fibrous intimal hyperplasia
Increase of fibrous and/or elastic matrix in the intima (scored as present or absent); arteries only
Accumulation of glassy, refractile strongly PAS-positive material in the intima and/or media (scored as present or absent); arterioles only
Fibrin thrombi in arterioles and/or glomeruli/(sub)total noninflammatory occlusion of the vessel by intraluminal, subendothelial, or medial accumulation of fragmented red blood cells, leukocytes, and eosinophilic material, and/or organized thrombi (scored as present or absent)
Cellular intimal hyperplasia
Proliferation of myofibroblasts in intima with reduction of the lumen of small-caliber arteries (scored as present or absent)
Focal cortical atrophy
Subcapsular ischemic cortical atrophy, sharply demarcated from normal (scored as present or absent, or cortex absent in section)
Table . HISTOPATHOLOGIC ITEMS FOR WHICH AGREEMENT AMONG 5 NEPHROPATHOLOGISTS WAS GOOD (INTRACLASS CORRELATION COEFFICIENT [ICC] >0.6) OR MODERATE (ICC 0.4–0.6)
Spikes were scored only if they covered >50% of the glomerular area.