Hepatocyte growth factor and transforming growth factor β1 ratio at baseline can predict early response to cyclophosphamide in systemic lupus erythematosus nephritis

Abstract

Objective

To determine whether the ratio of hepatocyte growth factor (HGF) to transforming growth factor β1 (TGFβ1) in systemic lupus erythematosus (SLE) nephritis could be a prognostic factor for response to therapy with cyclophosphamide (CYC) and steroids at 6 months, and to examine whether the molecular ratio of HGF to TGFβ1 correlates with the activity index (AI) and chronicity index (CI) and has predictive value for remission at the sixth month.

Methods

Thirty-six SLE patients with new-onset nephritis, 25 of whom were treated with CYC and steroids, entered into a prospective observational cohort trial at a tertiary university referral center. Renal biopsy findings and clinical parameters were recorded for all patients. Histopathologic, clinical, and immunohistochemical data at baseline served to define the predictive value for the outcome at 6 months.

Results

AI and CI at baseline did not distinguish patients who had achieved remission from those who had not achieved remission after receiving CYC plus steroids. HGF and TGFβ1 were expressed in the tubuli, not in the glomeruli. The CI correlated directly with the TGFβ1 extension score (TGFβ1-ES) (r = 0.43, P = 0.008), but correlated indirectly with the HGF intensity score (HGF-IS) (r = −0.39, P = 0.02) and the HGF-ES (r = −0.45, P = 0.006). An HGF-ES:TGFβ1-ES ratio of ≥1 at baseline distinguished patients who had achieved remission from those who had not achieved remission, with a predictive value of 94%.

Conclusion

These findings indicate that baseline expression of renal HGF and TGFβ1 predicts short-term renal outcome after therapy with CYC and steroids.

One-third of patients with systemic lupus erythematosus (SLE) nephritis have flares despite the best treatment and still progress to end-stage renal disease (1–4). Houssiau et al provided clear-cut data from a randomized controlled trial that response to therapy at 6 months was the best predictor of a good long-term renal outcome (5), thus suggesting that the first 6 months could be crucial for future renal function. Among the various other hypotheses, one might be that, at disease onset, kidney tissue biology and pathology could determine major outcomes, such as the response to standard treatment and unrelenting disease progression (6, 7), similar to the observation that erosions in rheumatoid arthritis and the biology of those erosions influence the future course of the disease and the response to treatment (8). Interstitial damage could be one of the factors determining disease outcome (9).

Studies of the role of hepatocyte growth factor (HGF) and transforming growth factor β1 (TGFβ1) in diabetic nephropathy have highlighted the fundamental role the balance between antifibrogenic/regenerative and profibrogenic/chronicity-inducing factors exerts in the final outcome of the disease (1). HGF plays an important part in tissue repair, proliferation, motility, and differentiation of renal tubular epithelial cells (10, 11). In animal models, TGFβ1 is a potent negative regulator of HGF expression, promotes extracellular matrix protein production, suppresses the proteolytic breakdown of extracellular matrix proteins, and induces epithelial cell growth arrest and apoptosis (12). We hypothesized that the balance between HGF and TGFβ1 expression in the first kidney biopsy of a patient with SLE nephritis could influence the outcome of the disease. We assessed these variables histologically and evaluated their relationship to the activity and chronicity of nephritis and their prognostic value for predicting clinically significant remission in patients receiving cyclophosphamide (CYC) and steroids for 6 months.

PATIENTS AND METHODS

Patients.

Thirty-six consecutive SLE patients with abrupt onset of renal disease who had been admitted to the Rheumatology Divisions of Rome and Udine Universities were enrolled in this study. All the patients met the American College of Rheumatology classification criteria for SLE (13) and gave their informed consent to participate to the study. For each patient, we evaluated the laboratory parameters of renal involvement (serum creatinine level, creatinine clearance, proteinuria) and immunologic features (antinuclear antibodies, anti–double-stranded DNA [anti-dsDNA], complement levels) at baseline and then every 3 months. A renal biopsy was performed at baseline in all patients, and after 2 years of therapy in 3 patients. At the time of the biopsy, all patients were receiving only low to moderate doses of glucocorticoids and/or antimalarial drugs.

Histopathologic and immunohistochemical analysis.

Specimens were fixed with phosphate buffered formalin, pH 7.2, for 4–18 hours, embedded in paraffin, sectioned at 3–4 μm, and stained with hematoxylin and eosin, Masson's trichrome stain, methenamine–silver, and periodic acid–Schiff. Only biopsy samples containing more than 7 glomeruli were considered.

All specimens were tested for immunofluorescence as follows. Paraffin-embedded sections (4 μm thick) were deparaffinized, rehydrated, and digested with Pronase solution (UCS Diagnostic, Rome, Italy), pH 7.5, at room temperature for 15 minutes. After washing in phosphate buffered saline (PBS), digested sections were incubated at room temperature for 1 hour with fluorescein isothiocyanate (FITC)–anti-IgG (1:30), FITC–anti-IgM (1:6), FITC–anti-IgA (1:6), FITC–anti-C3c (1:6), FITC–anti-C1q (1:6), and FITC-antifibrinogen (1:6) (all from Dako, Carpinteria, CA) diluted in PBS, pH 7.4. After washing, sections were mounted with Vectashield mounting medium (Vector, Burlingame, CA) and examined with a fluorescence microscope (DFC300 FX; Leica Microsystems, Milan, Italy).

Sections for immunohistochemical analysis were collected on 3-aminopropyltriethoxysilane (Sigma, Milan, Italy) or on naturally charged slides (Dako), allowed to dry overnight at 37°C to ensure optimal adhesion, dewaxed, rehydrated, and treated with 0.3% H₂O₂ in methanol for 10 minutes to block endogenous peroxidase. For antigen retrieval, sections were microwave-treated in 0.01M citrate buffer at pH 6.0 for 10 minutes and allowed to cool for 20 minutes. Endogenous biotin was saturated using a biotin blocking kit (Vector). Sections were incubated at room temperature for 25 minutes with rabbit anti-human HGFα (dilution 1:20, polyclonal; Assay Designs, Ann Arbor, MI) or mouse anti-human TGFβ1 (dilution 1:20, clone TGFB1; Novocastra, Newcastle, UK). Then the slides were incubated for 1 hour at room temperature with peroxidase-conjugated secondary antibodies, washed, incubated with diaminobenzidine, and counterstained with hematoxylin and eosin.

Negative controls for TGFβ1 were diseased tissue sections in which isotype-matched mouse Ig was used instead of anti-TGFβ1 (Figures 1A and B). In negative controls for HGF, preimmune rabbit serum instead of primary antibodies was used (Figures 1C and D). Positive and negative controls for HGF and TGFβ1 were normal liver and kidney tissue, respectively, obtained during surgical procedures performed on patients with cancer (Figures 1E–H).

Figure 1. A and B, Slides of adjacent sections of diseased renal tissue stained with transforming growth factor β1 (TGFβ1) (A) or treated with isotype-matched control immunoglobulin instead of TGFβ1 (B). C and D, Slides of adjacent sections of diseased tissue stained with hepatocyte growth factor (HGF) (C) or negative isotype-matched control (D). E and F, Sections of normal liver stained for HGF (E) or TGFβ1 (positive control) (F). G and H, Sections of normal kidney stained for HGF (G) or TGFβ1 (H). In some cells of the tubular brush borders, there are only a few small dots showing staining for TGFβ1. (Original magnification × 20.)

In each biopsy, the histologic class, according to World Health Organization (WHO) classification, and the activity index (AI) and chronicity index (CI) were evaluated (1). To quantitate the renal transcription of HGF and TGFβ1, sections of kidneys were examined by light microscopy and the degree of renal expression (intensity score [IS]) was defined as the mean value of the following immunohistochemical scores: 0 = no staining, 0.5 = trace staining, 1 = mild staining, 2 = moderate staining, and 3 = intense staining in at least 20 randomly chosen high-power fields (10). We also evaluated the extension of HGF and TGFβ1 transcription using the following scoring system defined as the extension score (ES): 0 = normal, 1 = involvement of <25% of the tissue field, 2 = involvement of 25–50% of the tissue field, 3 = involvement of 50–75% of the tissue field, and 4 = extensive involvement of >75% (14). The scores were evaluated by 2 different pathologists (GZ and VGV), working separately and under blinded conditions. All discrepancies were discussed and consensus reached. Furthermore, the immunohistochemistry scores were obtained independent of the histology indexes.

After the biopsy, 25 patients received pulses of steroids and pulses of CYC for 6 months according to the high-dose regimen described by Houssiau et al (5). All patients then received oral prednisone (0.5 mg/kg of body weight) daily for 4 weeks. The prednisone dosage was then tapered by 5 mg each week to the minimal dosage required to control extrarenal disease, or to 0.25 mg/kg/day. For severe extrarenal flares of lupus, patients were permitted to receive 1 mg/kg prednisone daily for 2 weeks (15) and the assessment of remission was made after 6 months. Eleven patients were treated according to different protocols (oral CYC for 3 months followed by cyclosporin A or azathioprine as part of a randomized trial). After 6 months, patients were divided into 2 groups: those who had achieved remission (proteinuria <1 gm/day and normal or decreased serum creatinine levels) and those who had not achieved remission (if one of the variables was not satisfied) (5).

Statistical analysis.

We calculated that for an 80% probability of success in identifying remission and with a 30% probability of treatment failure, with an alpha error of 0.05 and a power of 0.95, 23 patients would be needed for the study. Values are reported as the mean ± SD. Mean values were compared using nonparametric tests (Wilcoxon-Mann-Whitney) for continuous variables. Frequency tables were used to compare frequency distributions. Statistical analyses were performed with GraphPad Prism statistical software (GraphPad Software, San Diego, CA). P values less than 0.05 were considered significant.

RESULTS

Demographic characteristics of the 36 SLE patients are shown in Table 1. The patients included 29 women and 7 men, the mean ± SD age at first biopsy was 33.5 ± 10.9 years, and the mean ± SD SLE duration was 3.8 ± 4.0 years. Two kidney samples from each patient, containing 15 ± 5 glomeruli, small and medium blood vessels, interstitium, and tubules, were obtained by biopsy. Three patients had WHO grade II nephritis at renal biopsy, 3 had grade III, 24 had grade IV, and 6 patients had grade V nephritis. The mean ± SD AI score was 6.9 ± 3.9 (range 0–16) and the mean ± SD CI score was 2.9 ± 2.5 (range 0–9). Twenty-five patients (19 women, 6 men, mean ± SD age at first biopsy 34.5 ± 12.4 years, mean ± SD disease duration 3.0 ± 3.2 years, 2 patients with WHO grade II nephritis, 1 patient with grade III, 21 patients with grade IV, 1 patient with grade V) were treated with pulses of steroids and CYC (mean ± SD cumulative dose 6.2 ± 2.9 gm) (Table 1). Immunofluorescence analysis revealed immune deposits in all renal tissue samples (Table 2). In particular, findings from 18 biopsies were positive for IgM, 32 were positive for IgG, 15 for IgA, 18 for C3, 16 for C1q, and 8 for fibrinogen, in various combinations.

Table 1. Demographic and clinical characteristics of the patients at baseline*

	Total (n = 36)	Treated with intravenous CYC       (n = 25)
* Except where indicated otherwise, values are the mean ± SD. CYC = cyclophosphamide; anti-dsDNA = anti–double-stranded DNA; WHO = World Health Organization.
Age at time of biopsy, years	33.5 ± 10.9	34.5 ± 12.4
Disease duration, years	3.8 ± 4.0	3.0 ± 3.2
Sex, no. (%) female/no. (%) male	29 (80.5)/7 (19.5)	19 (76)/6 (24)
Caucasian, no.	36	25
Steroid dosage at biopsy, mg/day in prednisone equivalents	17.3 ± 14.7 (range 0–50, median 15)	18.2 ± 14.0 (range 0–50, median 15)
Anti-dsDNA positive, no. (%)	22 (61)	20 (80)
Serum creatinine, mg/dl	1.1 ± 0.6	1.1 ± 0.5
Creatinine clearance, ml/minute	71.1 ± 34.8	65.9 ± 23.7
Proteinuria, gm/24 hours	2.6 ± 1.6	2.5 ± 1.5
WHO grade of nephritis, no. (%)
I	0 (0)	0 (0)
II	3 (8.3)	2 (8)
III	3 (8.3)	1 (4)
IV	24 (66.7)	21 (84)
V	6 (16.6)	1 (4)
Activity index	6.9 ± 3.9	6.8 ± 4.8
Chronicity index	2.9 ± 2.5	3.6 ± 2.7

Table 2. Features of immune deposits in each patient in the cohort

Patient	IgM	IgG	IgA	C3	C1q	Fibrinogen
1	−	+	−	+	+	−
2	+	+	+	−	−	+
3	+	+	+	+	+	−
4	−	+	+	−	−	−
5	−	−	+	−	+	−
6	−	−	−	+	+	−
7	+	+	+	+	−	−
8	−	+	+	−	−	−
9	+	+	+	−	+	−
10	−	+	+	−	+	+
11	+	+	−	+	−	−
12	+	+	−	−	−	−
13	+	+	−	+	−	−
14	−	+	+	−	−	+
15	−	+	−	+	+	−
16	−	+	−	−	−	−
17	−	+	−	+	−	−
18	+	+	−	−	+	−
19	−	+	−	+	+	−
20	+	+	+	−	+	−
21	+	−	−	−	+	+
22	−	+	+	−	−	−
23	+	−	−	+	−	−
24	−	+	−	+	−	+
25	+	+	+	−	−	+
26	−	+	+	−	+	−
27	+	+	+	−	−	−
28	−	+	−	+	+	−
29	+	+	−	+	−	−
30	−	+	−	+	−	+
31	+	+	−	+	−	−
32	+	+	+	−	+	−
33	−	+	−	−	+	−
34	−	+	−	+	+	−
35	+	+	−	+	−	+
36	+	+	−	+	−	−

Figure 1 shows adjacent sections of diseased renal tissue stained with TGFβ1 (Figure 1A) and treated with isotype-matched control Ig instead of TGFβ1 (Figure 1B), and sections of diseased tissue stained with HGF (Figure 1C), and the negative control (Figure 1D). Immunohistochemistry analysis revealed that HGF and TGFβ1 were expressed in the tubuli, but not in the glomeruli (Figure 2).

Figure 2. A, Sections of kidney with low expression of hepatocyte growth factor (HGF) (extension score [ES] = 2) and high expression of transforming growth factor β1 (TGFβ1) (ES = 4). B, Sections of kidney with high expression of HGF (ES = 4) and low expression of TGFβ1 (ES = 2). (Original magnification × 2.5 in A; × 5 in B.)

We found a striking inverse correlation between the 2 molecules for the IS (r = −0.48, P = 0.003) as well as for the ES (r = −0.85, P < 0.0001). There was a highly significant negative correlation between the CI and the ratio of HGF to TGFβ1 (r = −0.44, P = 0.007). The CI also correlated directly with the TGFβ1-ES (r = 0.43, P = 0.008) (Figure 3A), but not with the TGFβ1-IS (r = 0.19, P not significant [NS]), and indirectly with the HGF-IS (r = −0.39, P = 0.02) as well as with the HGF-ES (r = −0.45, P = 0.006) (Figure 3B). No relationship was seen between the HGF-IS, HGF-ES, TGFβ1-IS, or TGFβ1-ES and any of the clinical or immunologic parameters of active nephritis (proteinuria, creatinine clearance, complement levels, anti-dsDNA antibody titer).

Figure 3. A, Significant correlation between chronicity index (CI) and transforming growth factor β1 extension score (TGFβ1-ES). No relationship was found between the intensity score (IS) of TGFβ1 and the CI. B, Significant inverse correlation between CI and hepatocyte growth factor IS (HGF-IS) and HGF-ES. Dots represent patients, but due to overlapping values, >1 patient may be represented by 1 dot.

After 6 months, 18 patients were defined as responders and 7 patients as nonresponders. In Table 3, the baseline characteristics of these patients are shown. Table 4 shows the serum creatinine levels and 24-hour proteinuria at baseline and at 6 months for each patient. Nonresponders were younger (all age ≤29 years), whereas other clinical parameters were similar in the responder and nonresponder subsets. HGF scores were significantly higher in the responders, while nonresponders had higher TGFβ1 scores. Moreover, the ratio of HGF-ES to TGFβ1-ES was significantly higher in the responder subset and, with a cutoff value of ≥1, the baseline ratio could be used to distinguish those who would achieve remission from those who would not (the ratio was ≥1 at baseline in 17 of 18 responders versus 1 of 7 nonresponders; P = 0.0002 by Fisher's exact test). The predictive value was 94%. No differences with regard to clinical and/or immunologic parameters, besides the younger age, were found in responders versus nonresponders.

Table 3. Parameters at baseline in 25 patients treated with CYC and steroids, by response group*

Variable	Responders (n = 18)	Nonresponders (n = 7)	P
* Except where indicated otherwise, values are the mean ± SD (range). CYC = cyclophosphamide; NS = not significant; HGF = hepatocyte growth factor; TGFβ1 = transforming growth factor β1; anti-dsDNA = anti–double-stranded DNA. † Median 15 mg/day in responders and 20 mg/day in nonresponders.
Age at first biopsy, years	38.3 ± 12.6 (16–67)	24.9 ± 3.2 (20–29)	0.003
Disease duration, years	3.2 ± 3.7 (0–13)	2.4 ± 1.9 (0–5)	NS
Steroid dosage at time of biopsy, mg/day in prednisone equivalents	15.6 ± 15.0 (0–40)†	23.8 ± 11.6 (15–50)†	NS
Activity index	6.9 ± 4.0 (2–16)	6.7 ± 5.6 (0–13)	NS
Chronicity index	2.6 ± 1.9 (0–6)	4.6 ± 3.5 (1–9)	NS
HGF extension score	3.3 ± 0.6 (2–4)	1.6 ± 0.5 (1–2)	0.0003
HGF intensity score	2.6 ± 0.6 (1–3)	1.3 ± 0.9 (0.5–3)	0.006
TGFβ1 extension score	1.3 ± 0.6 (1–3)	3.3 ± 0.75 (0.5–3)	0.0004
TGFβ1 intensity score	1.0 ± 0.4 (0.5–2)	2.1 ± 0.9 (1–3)	0.008
HGF:TGFβ1 intensity ratio	2.9 ± 1.3 (0.5–3)	0.8 ± 0.9 (0.5–3)	0.003
HGF:TGFβ1 extension ratio	3.1 ± 1.1 (1–4)	0.5 ± 0.3 (0–1)	0.0004
Proteinuria, gm/24 hours	2.4 ± 1.5 (0.1–7.4)	2.7 ± 1.5 (0.4–4.5)	NS
Serum creatinine, mg/dl	1.1 ± 0.6 (0.5–2.9)	1.0 ± 0.5 (0.5–2.1)	NS
Creatinine clearance, ml/minute	60.7 ± 8.6 (16–173)	71.2 ± 38.8 (53–80)	NS
C3, mg/dl	61.3 ± 25.5 (15–101)	63.2 ± 26.8 (53–90)	NS
C4, mg/dl	9.3 ± 3.1 (4–29)	11.4 ± 6.6 (6–10)	NS
Anti-dsDNA, no. (%) positive	14 (77.7)	6 (85.7)	NS

Table 4. Serum creatinine levels and 24-hour proteinuria values in each patient at baseline and at 6 months

Patient	24-hour proteinuria, gm		Serum creatinine, mg/dl
	Baseline	6 months	Baseline	6 months
Responders
1	1.5	0.8	0.6	0.6
2	2.5	0.1	1.4	1.0
3	2.0	0.8	0.8	0.7
4	2.2	0.7	0.9	1.0
5	3.3	0.9	0.9	0.7
6	2.8	0.6	1.2	0.8
7	3.0	0.9	1.0	0.6
8	3.2	0.8	1.1	0.8
9	1.2	0.5	1.5	1.1
10	3.2	0.8	2.9	1.1
11	7.4	0.2	0.8	0.5
12	1.4	0.7	0.5	0.6
13	2.2	0.8	0.9	1.0
14	2.0	0.6	1.2	0.9
15	3.5	0.5	1.7	0.9
16	1.4	0.5	1.4	0.8
17	0.1	0.0	1.5	0.8
18	1.2	0.4	0.8	0.5
Nonresponders
19	1.8	4.0	1.0	0.7
20	4.5	5.0	1.0	3.6
21	0.4	1.0	1.0	1.3
22	3.6	2.8	0.5	0.7
23	1.5	1.2	1.0	0.9
24	3.9	3.1	0.7	0.5
25	3.5	2.6	2.1	7.4

Nine of the 25 patients (5 responders and 4 nonresponders) were taking angiotensin-converting enzyme (ACE) inhibitors. No differences in any parameters (particularly proteinuria, serum creatinine level, and HGF and TGF values) were found between those taking and those not taking ACE inhibitors.

Fourteen of the 25 patients had hypertension (10 of the 18 responders and 4 of the 7 nonresponders). No differences in any parameters were found between patients with and those without hypertension.

The mean ± SD daily dosage of steroids being taken at the time of biopsy was 15.6 ± 15.0 mg prednisone equivalents in the responder group and 23.8 ± 11.6 mg in the nonresponder group. Analysis showed that the baseline scores of HGF and TGFβ1 were not different when comparing patients taking ≤15 mg/day steroids versus >15 mg/day (HGF-ES 3.2 ± 0.8 versus 2.5 ± 1.3, P NS; TGFβ1-ES 1.5 ± 0.8 versus 2.3 ± 1.3, P NS; HGF-IS 2.4 ± 0.7 versus 2.0 ± 1.1, P NS; TGFβ1-IS 1.3 ± 0.8 versus 1.4 ± 0.8, P NS).

To investigate how therapy could modify the HGF and TGFβ1 scores, we assessed the kidneys of 3 patients at baseline and after 2 years of cyclosporin A therapy. The renal function levels of the 3 patients were as follows: creatinine clearance (corrected for body surface area) 71.6, 79, and 79.1 ml/minute, respectively, at baseline and 94.6, 70.4, and 95 ml/minute, respectively, at 2 years; proteinuria 0.5, 1.6, and 3.2 gm/24 hours, respectively, at baseline and 0.1, 0.4, and 0.2 gm/24 hours, respectively, at the time of the second kidney biopsy. In these patients, the second biopsy showed an increase in TGFβ1 expression compared with the basal level (the HGF score in the 3 patients was 4, 1, and 4, respectively, at baseline and 3, 2, and 3 after 2 years; the TGFβ1 score was 1, 3, and 2, respectively at baseline and 2, 3, and 3 after 2 years). The cutoff value of ≥1 for the baseline ratio of HGF-ES to TGFβ1-ES was predictive of the medium-term course of the disease in these patients, but this ratio changed in the long term with cyclosporin A therapy.

DISCUSSION

In a recent prospective study of SLE nephritis, Houssiau et al (5) found that early (6 month) response to therapy (defined as a normal serum creatinine level and proteinuria of <1 gm/24 hours) was the best predictor of good long-term renal outcome. Yoo et al (16) and Leaker et al (17) had observed that a high histologic glomerular activity index was predictive of poor renal function, but Hill et al (6, 7) reported that the initial biopsy had very little predictive ability, whereas findings on the second biopsy, performed after 6 months of therapy, could be used to accurately predict future kidney function. Consistent with Hill et al's studies, we also found no histopathologic variables that were at baseline, even though the CI tended to be higher in nonresponders. The AI and CI at baseline do not seem to be sensitive enough to discriminate between those who will have a poor versus a favorable outcome (18).

The role of interstitial and tubular damage has received attention in the past. Cameron (19) observed that there was a strict relationship between intensity of interstitial infiltration and impairment of renal function and subsequent rate of progression of renal disease. In subsequent studies analyzing the tubulointerstitial activity index and the relationship between interstitial inflammation and the chronic lesions index, Hill and colleagues showed that the interstitial variables (tubular damage variables and interstitial inflammation) had a strong correlation with chronicity components, and with serum creatinine levels (20, 21). Esdaile et al (22) developed a tubulointerstitial index involving tubular atrophy, interstitial fibrosis, and interstitial inflammation, and concluded that this was the best overall predictor of outcome. D'Amico (23), from the evidence in the multivariate analysis in various studies of the 3 most common glomerular diseases (IgA nephropathy, membranous nephropathy, and type 1 membranoproliferative glomerulonephritis), concluded that tubulointerstitial damage was the strongest predictor of a subsequent poor outcome, and that TGFβ1, as the final driver of several pathways, was the most likely key molecule.

Our study provides data to support the value of early biopsy with evaluation of the ratio of HGF to TGFβ1, to predict the response to CYC and steroids at 6 months, in patients with lupus nephritis. HGF and TGFβ1 were inversely transcribed in the kidneys of SLE patients. HGF levels correlated inversely with TGFβ1 levels, and the ratio of HGF-ES to TGFβ1-ES was inversely related to the CI. In addition, our study suggests the importance of baseline tubular damage as a prognostic factor for response to standard therapy. Specifically, tubular damage assessed by the ratio of HGF to TGFβ1 was strongly predictive of the response to CYC at 6 months.

Nephritis in SLE is one of the most important causes of morbidity and mortality, and pathologic diagnosis and timing of therapeutic intervention could become the crucial prognostic variables. Ciruelo et al (4) reported that a delay of >5 months from the onset of nephritis to initiation of CYC therapy was a determinant of a poor early response and of a poor long-term result. Our data indirectly support these clinical data, since TGFβ1 transcription reflects ongoing chronicity. In the few patients in whom another biopsy was obtained after 2 years of cyclosporin A therapy, we observed, as expected, increased expression of TGFβ1 and decreased expression of HGF. Even in these cases, the baseline ratio of HGF to TGFβ1 predicted the course of the disease. We also confirmed the previously reported finding that younger age at renal biopsy is a predictor of poor outcome (3).

Further independent studies are needed to confirm our findings, especially in patients whose nephritis is the first manifestation of the chronic disease. If the prognostic value of the HGF to TGFβ1 ratio is confirmed even in these cases, the notion that the interstitium and tubular damage play a key role would be strongly reinforced.

The observation that a high HGF score and a low TGFβ1 score heralds a favorable response to CYC and steroids in SLE nephritis, while a low HGF score and a high TGFβ1 score almost uniformly predicts a poor response, provides strong evidence that therapeutic strategies to induce, at an early stage, higher HGF gene expression and/or suppression of TGFβ1 gene function would be of benefit in the treatment of this disease.

Acknowledgements

The authors thank Professor Arnaldo Capelli and Dr. Vincenzo Arena (Institute of Pathology, Catholic University, Rome) for helpful advice in the preparation and interpretation of the positive and isotype-matched negative control tissue slides.