OUP user menu

Henoch Schönlein purpura with nephritis in adults: adverse prognostic indicators in a UK population

S. Shrestha, N. Sumingan, J. Tan, H. Althous, L. McWilliam, F. Ballardie
DOI: http://dx.doi.org/10.1093/qjmed/hcl034 253-265 First published online: 24 March 2006


Background: Henoch Schönlein purpura with nephritis (HSN) in adults may cause severe organ injury, but its rarity has contributed to a lack of data.

Aim: To evaluate clinical outcomes and risk factors in adult HSN patients.

Design: Retrospective analysis.

Methods: Thirty-seven patients with adult HSN attending the Regional Vasculitis Clinic between 1974 and 2004 were assessed. For inclusion, a renal biopsy showing predominant mesangial IgA immune deposits was required, plus at least two of: purpuric rash, arthralgia, abdominal pain.

Results: Ten patients (27%) progressed to end-stage renal failure (ESRF). Renal failure rates were highest in the first decade, with survival rate 72% at 5 years, 68% at 10 years and 46% at final review. Risk factors for ESRF were: proteinuria ⩾1 g/day during follow-up (RR 83.8, p = 0.0006); hypertension at presentation (RR = 53.3, p = 0.0045) and during follow-up (RR = 5.9, p = 0.05); renal impairment at presentation (RR 8.0, p = 0.0015); age <30 years (RR 7.6, p = 0.02); and male sex (RR = 6.0, p = 0.05). Biopsies frequently showed crescents, mostly affecting <50% of glomeruli; their presence predicted ESRF, as did interstitial fibrosis and tubular atrophy. Renal remission, in contrast, was also high (43%). Cytotoxics were used in 32%, with no clear effect on outcome. Relapses affecting the classical extra-renal systems were common, but were not associated with declines in renal function. A high proportion of patients (41%) also suffered vasculitic organ injuries outside the classical systems.

Discussion: HSN in adults is a serious relapsing disease, causing renal failure as frequently as in small-vessel ANCA-positive vasculitides. Prognosis and risks differed in this series from those in other countries, including a higher risk of ESRF than in previous series. Distinct groups developed either ESRF, or remitted. The absence of clear benefit suggests that corticosteroids should be reserved for patients with serious disease, and that cytotoxics may not be merited for those at high risk of renal failure.


Henoch Schönlein purpura (HSP) with nephritis (HSN) in adults may involve severe organ injury (Figure 1), but is infrequently seen in acute general internal medicine and specialty practice. There is a near 20-fold disparity in incidence between adults and children (1.31 vs. 22.12 per 100 000 population, respectively), which may partly explain the paucity of data on adults. No previous UK reports have enabled comparison with contemporary series in other countries.3–5 The historical description of HSP as a relapsing disease after Heberden's initial writings in 1802, was first offered by Johann Schönlein in 1834—' … the (only) exanthem that never happens all at once, but in bouts … '—but in contrast to other vasculitides, there is scant description of its natural history in adults.

Figure 1.

a Necrotizing bullous vasculitis in an adult male with acute Henoch Schönlein purpura. b Segmental necrotizing crescentic glomerulonephritis with c characteristic mesangial immunoglobulin A deposits.

Identification of cases may have been impeded by the American College of Rheumatology HSP criteria,6 which restricted diagnosis to those aged <20 years, though there were attempts at revision by the CHCC7 and by Michel in 1994.8 The classic tetrad of cutaneous vasculitis, arthralgias, gastrointestinal vasculitis, and nephritis, three of which suffice for diagnosis, is now generally accepted only when immunoglobulin A is demonstrable in an affected organ, to enable clear distinction from other forms of small-vessel vasculitis presenting with similar clinical features. In practice, this means CHCC criteria, using glomerular mesangium to identify immunoglobulin A deposits, because of inconsistencies in reproducing skin biopsy immunofluorescence.9 Renal biopsy has become an essential investigation to avoid these false negatives, and to distinguish HSP from other vasculitides in adults. For these reasons, HSN now accounts for almost all firm diagnoses of HSP in adults, as a diagnosis of HSP without nephritis will be equivocal if skin biopsy is negative for immunoglobulin A. Thus HSN in adults is, in effect, synonymous with the historical diagnosis, HSP.

After the initial acute organ injuries (when gastrointestinal vasculitis may threaten survival) have settled, renal disease is the main determinant of morbidity. In some populations, such renal disease is more frequent in adults than in children: Italy, 32% in adults vs. 25% in children10 at 5 years; Spain, 32% in adults (at 5 years 10%) vs. 0% in children;5 but 8% in both in Singapore.11 ESRF at 5 years was 15% in adults vs. 5% in children in Italy, but at 10 years, it was 16% in both.10 Nevertheless, the perception remains that prognosis is worse in adults. Risks of developing renal insufficiency would appear to range from 5% to 15%12,,13 in children, and from 0% to 49% in adults,13–19 with the prevalence of 32% at 15 years in the recent French study4 being an approximate average.

Only seven literature studies include the necessary diagnostic criterion of immunofluorescence or histology. These evaluated clinicopathological variables in adults,3–5,10,11,16,,17 primarily histology (all), proteinuria (in five), and hypertension (in three), but even the most recent lacks blood pressure data. Now that target pressures of 125/75 mmHg20,,21 have been accepted, the importance of providing these data is obvious, when analysing pathophysiological determinants of prognosis. Hypertension was an adverse factor only in the Italian study,10 not in the others.3,,4 Proteinuria at presentation >1 g/24 h is an adverse determinant, one study excepted,16 consistent with findings in IgA nephropathy (reviewed in reference 22), but proteinuria during follow-up has not been rigorously assessed. In contrast to intense efforts to find serological markers of prognosis in IgA disease, there have been no equivalents in HSN. Only one study has suggested that raised serum immunoglobulin A is associated with a favourable prognosis,16 another contending that this finding was age-related only.4

Data from these UK adult HSN patients were analysed for factors associated with adverse prognosis and organ injuries during the natural history of the disease.


We retrospectively assessed 37 patients with HSN aged >16 years at disease onset, attending the Regional Vasculitis Clinic in a single teaching hospital centre, between 1974 and 2004. The diagnosis was established using CHCC criteria, with our suggested modification of restricting identification of immunoglobulin A to the glomerular mesangium. Thus at presentation, all patients had renal disease (significant erythrocyturia, 24 h urine protein and/or serum creatinine >120 μmol/l) and at least two of: purpuric rash, arthralgia, abdominal pain.

Erythrocyturia, both initially and during follow-up, was estimated from a timed 2-h morning urine specimen. Microscopy was done in a defined volume counting chamber, using a method analogous to the Addis count23 to determine significant microscopic haematuria (>1 × 106/24 h for males, and >2 × 106/24 h for females) for an uncontaminated specimen.24,,25 Data were collected on the demographic, clinical and other laboratory findings of the study patients. Patient referrers were identified, and a communication network set up with dermatologists and acute medicine physicians (the main referrers) to facilitate simple urine screening on patients with cutaneous vasculitis. Recurrences of purpura, haematuria, gastrointestinal and joint symptoms, as well as atypical organ injuries, were recorded during follow-up.

The primary end-point was ESRF; patient data were censored if lost to follow-up or if death occurred during the disease course. For study inclusion, renal biopsies were required to show predominant mesangial IgA immune deposits by immunofluorescence; two patients suspected of having HSN, but who did not fulfil this criterion, were excluded from further analysis, as were those fulfilling the American College of Rheumatology26 and Chapel Hill Consensus Conference27 criteria for other primary systemic vasculitides, or those with other systemic autoimmune diseases and cutaneous vasculitis.

Clinical definitions

Skin involvement was defined as non-thrombocytopenic palpable purpura, typical distribution extensor surfaces of the lower limbs and buttocks; joint involvement as symptomatic arthralgia and/or evidence of synovitis; gastrointestinal involvement, if there was a complaint of colicky abdominal pain accompanied by change of bowel habit or with evidence of bleeding.

Hypertension was defined as systolic blood pressure >130 mmHg and/or diastolic blood pressure >80 mmHg and/or patients receiving any antihypertensive medications, documenting use of ACE inhibitors and angiotensin-II-receptor-blockers. Longitudinal data on efficacy of blood pressure control was also abstracted. Proteinuria (measured on one or more occasions) was stratified by mean value (g/24 h): mild, 0.3 to <1; moderate, ⩾1 to <3.5; severe, ⩾3.5. Haematuria was classified by erythrocyte count per 24h (microhaematuria, ⩾106 but <108; macrohaematuria, ⩾108) or, for purposes of clinical analyses, if unequivocal description by the patient was recorded. Renal status at the last visit was recorded if the patient was lost to follow-up or underwent dialysis, and similarly for mortality data, together with the cause of death.

Disease relapse was recorded when a patient previously diagnosed with HSN who had been asymptomatic for at least 1 month, presented with a new flare of cutaneous vasculitis or other systemic involvement attributable to HSP. Outcome was classified as remission when, in the absence of extra-renal involvement, serum creatinine was stable (at <130 μmol/l in males, or <120 μmol/l in females) without significant proteinuria or haematuria in two or more clinic assessments, and over a minimum period of 6 months.

Renal biopsy morphology criteria

Renal biopsy data from specimens obtained within 1 month of presentation were evaluated by a single pathologist blinded to clinical information, using criteria potentially associated with prognosis, as previously summarized.28 All biopsies were stained with haematoxylin and eosin (H&E), periodic acid schiff (PAS) and methenamine silver, and were analysed and scored semi-quantitatively (0–3) for the following features: (i) percentage of glomeruli showing global sclerosis; (ii) percentage affected by segmental and global glomerulosclerosis; (iii) percentage with crescents; (iv) severity of mesangial cell proliferation (each glomerulus scored individually, mean score calculated for all non-sclerosed glomeruli); (v) severity of increase in mesangial matrix; (vi) extent of interstitial fibrosis and tubular atrophy; (vii) severity of arterial/arteriolo-sclerosis or hyalinosis.

Percentage measures (i–iii) were scored as: none; 1, ⩽24%; 2, 25–50%; 3, >50%. The others were scored as: 0, none; 1, mild; 2, moderate; 3, severe. A biopsy containing eight or more glomeruli was taken as adequate for histological analysis.

Statistical analyses

Results were expressed as numbers (%) for categorical variables and as range or mean ± SD for continuous variables. Comparisons used the χ2 test, or Fisher's exact test if expected value was <5. The Wilcoxon Mann–Whitney test was used when data were unlikely to be normally distributed. Kaplan–Meier functions were used for survival curve calculations. The following variables were used: age, gender, hypertension, renal impairment, proteinuria, degree of proteinuria, microhaematuria, and macrohaematuria. Variables were divided into two groups: those at onset, and during follow-up. Multivariate analysis was not performed because of subpopulation numbers. Tests were two-tailed where appropriate, and significance levels were taken as 5%. Statistical analyses used SPSS (v. 11.5).


Patient characteristics

Of the 37 HSN patients, 22 (59%) were male and 15 (41%) female (M:F 1.5:1). Age of onset was 16–67 years, mean 35 ± 17. Duration of follow-up was 9.4 ± 7.1 years, range 2–30. Most patients were referred from secondary care: acute medicine 35%, dermatology 33%, rheumatology and other specialties 19%, primary care 13%.

Organ injuries

Aspects of the natural history of HSP are illustrated in Figure 2.

Figure 2.

Disease expression during the course of adult Henoch Schönlein purpura. Disease remained active, with clinical organ involvement in a minority of affected adults after a minimum 2-year follow-up.

(i) At disease onset

Seventeen patients (46%) had histories of infection preceding presentation (8 upper respiratory tract, 4 tonsillitis, 2 chest infections, 3 other viral illnesses). Insect bites were reported in a further two. At onset, all had palpable purpura. Significant proteinuria and arthralgias were present in 31 (84%). All had urinary abnormalities (proteinuria and/or haematuria) at diagnosis; 25 (68%) had moderate or severe proteinuria (>1 g/24 h). Macroscopic haematuria was reported in 18 (49%), but did not confer protection against renal failure (unlike in the disease counterpart, IgA nephropathy). Microscopic haematuria was found in 46%, and 54% had gastrointestinal involvement. Renal function was impaired in 32%, and 59% were hypertensive.

Other, non-classical, system involvement was frequent (41%): 11 (30%) reported epistaxis; three (8%) had eye involvement (episcleritis, impaired right visual field, swollen eyelids); one had neural involvement (left common peroneal nerve palsy) and one had haemoptysis. In contrast to the findings of other recent studies,4,5,,14 we did not find age-related differences in clinical presentation (Table 1). Laboratory findings were generally unhelpful: elevated IgA was found only in four patients (11%), lower than in most reports of IgA nephropathy. Twenty patients (54%) had a raised ESR at presentation.

View this table:
Table 1

Clinical features of adults with Henoch Schönlein nephritis at presentation: absence of age-related differences, in contrast to other series

Clinical variableAge <30 years (n = 18)Age ⩾30 years (n = 19)p
Total numbers (%)
Recent history of infection8 (44%)9 (47%)0.89
17 (46%)
Purpura18 (100%)19 (100%)1.0
37 (100%)
Proteinuria15 (83%)16 (84%)0.72
31 (84%)
Microhaematuria9 (50%)8 (42%)0.32
17 (46%)
Macrohaematuria6 (33%)12 (63%)0.32
18 (49%)
Joint involvement14 (78%)17 (89%)0.37
31 (35%)
Gastrointestinal involvement12 (67%)8 (42%)0.09
20 (54%)
Renal impairment7 (39%)5 (26%)0.86
12 (32%)
Hypertension10 (56%)12 (63%)0.52
22 (59%)

(ii) During follow-up

Classical extra-renal organ disease, both episodic and persisting, was seen in 17–43% of patients during follow-up and in 5–14% at final review (minimum 2 years) (Figure 2). There was evidence of persisting nephritis, proteinuria (60%) and/or haematuria in over two-thirds of the patients during follow-up, and in at least 20% at final review. Microscopic haematuria increased in prevalence during follow-up, from 46% at presentation, to >60% at final review. Non-classical extra-renal system disease resolved in affected cases. In contrast to presentation, relapses were not generally preceded by a history of infective illness.


At final review, 43% were in clinical remission, 27% had progressed to ESRF, a further 8% had impaired renal function and 19% had persistent proteinuria. The time interval between disease onset and ESRF ranged from 3 months to 21 years, median 7.0 years. Of the 10 who progressed to ESRF, two (20%) had rapidly progressive glomerulonephritis, losing function in a period of 3 months. Most (70%) of the patients developing ESRF did so in the interval 5–8 years after disease onset, but renal failure continued to develop two decades after presentation (Figure 3). Five deaths (14%) were reported by the end of follow-up. One patient died after complications of dialysis; three deaths resulted from asthma, neoplasia or bronchopneumonia, and the cause of death was unspecified in one.

Figure 3.

Kaplan–Meier analysis showing cumulative renal survival for all Henoch Schönlein purpura patients. Renal survivals for patients with mean proteinuria >1 g/24 h during follow-up vs. those with lesser proteinuria (p = 0.001).

Risk factors for ESRF

(i) Proteinuria and haematuria

Renal survival differed according to clinical variables at presentation (Table 2). In proteinuric patients, renal survival was 40% at 10 years, and 25% at 20 years, vs. 100% in non-proteinuric patients (data not shown; p = 0.0006). The severity of proteinuria during follow-up (Figure 3), not at presentation, subdivided this group: all 10 patients who progressed to ESRF had moderate/severe proteinuria (⩾1 g/24 h) during follow-up. Similarly, 59% of those with proteinuria ⩾1 g/24 h during follow-up progressed to ESRF, vs. none of those with proteinuria <1 g/24 h (p = 0.001). Moderate/severe proteinuria during follow-up was thus a highly significant predictor of renal survival, but proteinuria at presentation, even severe, had no such influence. Macroscopic haematuria had no effect on renal survival (Table 2).

View this table:
Table 2

Henoch Schönlein purpura in adults: renal outcome according to clinical status at presentation or follow-up

VariableNormal renal function (creatinine <130 μmol/l) (n = 24)ESRF (n = 10)p
Proteinuria ⩾1 g at presentation15 (63%)8 (80%)0.43
Proteinuria (mean value) ⩾1 g during follow-up4 (17%)10 (100%)0.001*
Renal impairment at presentation2 (8%)8 (80%)0.0015*
Hypertension (presentation)10 (42%)10 (100%)0.002*
Hypertension (follow-up)12 (50%)9 (90%)0.05*
Macroscopic haematuria15 (63%)6 (60%)1.0
Mean ± SD age (years)38 ± 16.221 ± 8.50.004*
  • *Significant values.

(ii) Renal impairment at disease onset

Renal impairment at presentation was also a prognostic marker: 8/10 (80%) progressed to ESRF vs. only 2/24 (8%) with initially normal renal function. Mean time to ESRF was 6.5 years (Table 2). Ten-year renal survival was 39% in those presenting with renal impairment vs. 88% in those with initially normal renal function (p = 0.0015, Figure 4).

Figure 4.

Renal survival after presentation with patients with a initially normal renal function vs. b impaired renal function (p = 0.0015).

(iii) Hypertension

Renal outcome was favourable in all normotensive patients. Hypertension at presentation and during follow-up (treated) were both associated with adverse renal outcome (p = 0.0045 and 0.05, respectively) (Table 2). In all hypertensives, renal survival was 45% at 10 years, and 22% at 20 years, contrasting with 92% in normotensive patients (Figure 5). ESRF progression was 45% in those hypertensive at presentation vs. 0% in normotensives.

Figure 5.

Renal survival in all patients: a hypertensive vs. b normotensive (p < 0.005).

Mean arterial pressures achieved during follow-up were higher than the recent widely accepted targets of 125/75 mmHg.20,,21 However, pressures were higher at presentation only in those progressing to ESRF vs. those who did not; at 3 years, pressures were somewhat lower in the former group. Throughout most of follow-up, however, there was no significant difference between pressures in the two groups (Figure 6; Wilcoxon Mann–Whitney). Of the 37 patients, 15 received ACE inhibitors or angiotensin-II-receptor-blockers; of these 15, six developed ESRF, one developed impaired renal function, and eight retained renal function within the normal range. Of the 22 patients not receiving either drug type, four progressed to ESRF, two developed impaired function, and sixteen retained normal renal function (p = NS, χ2 test).

Figure 6.

Mean arterial pressure in ESRF group vs. non-ESRF group. Pressures differed only at presentation (p = 0.04), and one occasion during follow-up (at 3 years) (p = 0.01). Wilcoxon Mann–Whitney tests.

(iv) Age and sex

Renal survival in those aged <30 years was poor compared to those aged >30: more than 75% of younger patients progressed to ESRF over 20 years (p = 0.02, Figure 7). Of the 10 patients who progressed to ESRF, nine were male and one female. Cumulative renal survival at 10 years was 57% in males vs. 89% in females (p = 0.05, Figure 8).

Figure 7.

Renal survival in a patients aged ⩾30 years vs. b aged 16–29 years (p = 0.02).

Figure 8.

Renal survival in a females vs. b males (p = 0.05).

(v) Histology

In 34/37 patients for whom biopsy data were available, the degree of interstitial fibrosis and tubular atrophy initially present were the only histological variables enabling differentiation of those retaining normal renal function from those progressing to ESRF (p = 0.005, Figure 9a). Data from the three patients developing impaired renal function were omitted from this analysis for clarity of presentation. Crescent formation was found in 12 patients (33%), (26% of those retaining normal renal function) at scores approaching 50% (Figure 9b). Crescents were also found in half of those progressing to ESRF (χ2 = 12.2; p = 0.0005). Thus, despite the high proportion found in those retaining renal function, their presence was an adverse prognostic finding.

Figure 9.

a Histological findings in renal biopsies at presentation (mean ± SEM) (p = 0.005, Wilcoxon Mann–Whitney test). b Crescent score data: 33% of all biopsies had one or more cellular crescents, including 50% of those progressing to ESRF, but also 26% of those retaining normal long-term renal function (p = 0.005; q.v. text).

Prediction of renal outcome by univariate analysis

We used univariate analysis to assess the relative risk (RR) for developing ESRF of several covariates: age, renal impairment at presentation, proteinuria at onset, proteinuria during follow-up, hypertension at onset, and hypertension during follow-up (Table 3). Moderate or severe proteinuria during follow-up was the strongest predictor of ESRF (RR 83.8, p = 0.0006), followed by hypertension at presentation (RR = 53.3, p = 0.0045), renal impairment at presentation (RR 8.0, p = 0.0015), age <30 years (RR 7.6, p = 0.02), male sex (RR = 6.0, p = 0.05), and hypertension during follow-up (RR = 5.9, p = 0.05). The small numbers of patients in subgroups precluded multivariate analysis.

View this table:
Table 3

Prognostic risk factors for developing end-stage renal failure, by univariate analysis

Age <30 years7.60.96–60.20.02*
Renal impairment at presentation8.01.7–37.00.0015*
Proteinuria ⩾1 g/24 h at presentation1.70.4–8.30.48
Proteinuria ⩾1 g/24 h during follow-up83.80.37–18 966.00.0006*
Hypertension at presentation53.30.30–9332.90.0045*
Hypertension during follow-up5.90.75–46.90.05*
Macrohaematuria at presentation1.40.41–5.20.55
Male sex6.00.7–48.00.05*
  • RR, relative risk. *Significant value.

Immunosuppressive treatment

Drug therapy was given to patients with progressive renal disease, and to those who presented with severe clinical and histological features. Corticosteroids were prescribed to 22% for severe gastrointestinal vasculitis, and cyclophosphamide added when there were severe histological renal biopsy findings and/or rapidly progressing renal disease, in a further 32%. Of the 10 ESRF patients, 60% had received prednisolone and cyclophosphamide. Eight patients from the group retaining normal renal function had received prednisolone. There were no significant differences in the outcome between those who received these agents and those who did not. The small number of patients and the retrospective nature of this study did not permit the different treatment protocols used to be compared or analysed to a satisfactory evidence-base standard. Nine ESRF patients received kidney transplants; none lost their grafts as a result of recurrence of HSN.

Disease relapses

No significant difference in renal outcome was observed between patients who had relapses in extra-renal organs vs. those who did not. Relapses did not appear to be associated with an acute decline in renal function. Purpura was the commonest form of relapse at 41% of events, with gastrointestinal symptoms in 16%, and arthralgias in 11%. A history of infection was generally absent in relapses.


This UK series describes the natural history of organ disease and long-term proteinuria in adults with HSN. The likelihood of remission, and the risk factors leading to morbidity and ESRF, are also evaluated.

The frequency of ESRF (27%) is the highest yet described in adults with HSN, and almost double that of the next highest (16% in Italy10).The renal attrition rates (54% at completion of follow-up, 28% at 5 years, 32% at 10 years, Kaplan–Meier analysis) are also the highest reported. Although follow-up was twice as long here (mean 9.4 years) as in the Italian study, which might explain a higher frequency of ESRF, the recent French study, with 14.8 years follow-up, reported only 11% ESRF,4 and there are similar results in other populations (Table 4). Strict diagnostic criteria were used throughout, similar to those of the French study, so these findings are likely to be substantive. Because of disease rarity, most series have been small, or with brief follow-up, making reliable analysis of prognostic factors difficult.11,16,17,,29 Numbers in this study, the second largest series in the last decade, still permitted only univariate analysis, but the longitudinal data on proteinuria and clinical variables are robust and comprehensive.

View this table:
Table 4

Summary of previously published studies on adult Henoch Schönlein purpura

Year of publication19861989199719972002200220022006
Follow-up (years) (median* or mean)3.2*7.5*1.254.9*6.56.1*14.87.0
Age (years) (mean)27.236.753.227.544.8425035
Gender (M:F)
Clinical presentation (%)
Renal features:
    Renal impairmentb29b31b13e21ND34g32i32j
Renal outcome (%)
    Renal impairment050932b1011388
    End-stage renal failure0190.0167111127
  • aProtein excreted >1 g per 24 h, at presentation only; bSerum creatinine >1.5 mg/dl (upper normal 1.4 mg/dl); cBlood pressure >140/90 mmHg; dBlood pressure >150/90 mmHg; eSerum creatinine >125% of the upper limit of normal; fProteinuria minimal (⩽1 g/24 h) or moderate (>1 but <3.5 g/24 h); gCreatinine clearance ⩽85 ml/min; hUse of antihypertensive agents or systolic BP >140 mmHg and/or diastolic BP >90 mmHg; iCreatinine clearance <50 ml/min (calculated by Cockroft–Gault formula); jSerum creatinine >130 mmol/l; kUse of antihypertensive agents or systolic BP >130 mmHg and or diastolic BP >80 mmHg; ND, no data.

Long-term mortality and morbidity in this study were attributable to complications of renal disease; more serious extra-renal features (gastrointestinal) subsided early. In contrast, a study of nephrotic-range proteinuria HSN in children30 suggested that organ injury at presentation was as important as biopsy grade, since these were more predictive of outcome than severity of glomerulonephritis. Deaths were low in this series (14%), approximately half that reported in the larger adult HSN multicentre study from France (26%).4 There was a longer follow-up in that study, but this UK series had a higher frequency of renal failure, so might reasonably be expected to have higher, rather than lower, mortality.

In contrast to the high frequency of renal failure, the rate of renal remission was also high (43%). In the French study, renal function was normal in 68% at presentation, but remission followed in only 20%. Our patients thus present a more polarized picture, with distinct groups at greater risk of developing renal failure, but others more prone to renal remission long-term.

Extra-renal disease relapsed more frequently here than in other series, with recurrent rather than single4 episodes in up to 50%, mainly in the classical organs (Figure 1). At completion of follow-up (minimum 2 years), a minority of adults (<20%) still had active extra-renal disease: cutaneous more often than gastrointestinal, but milder than at presentation, not resulting in ulceration or haemorrhage, respectively. This contrasts with expectations in children.

Patients in this series presented during similar historical periods to those of other studies, so the results may represent a genuine difference of some kind, whether in genetics, environment, prevalence, effectiveness in control of hypertension, or immunosuppressive drug use. Our use of corticosteroids appears similar to that of the French study4 (54% vs. 56%), but our use of cytotoxics (cyclophosphamide) was twice as prevalent (32% vs. 17%), consistent with the more severe renal injuries in our series. However, use of immunosuppressive drugs did not appear to influence prognosis in this series, nor did use of ACE inhibitors or angiotensin-II-receptor-blockers. Therapy selection, as in other series, precluded firm conclusions on improving the therapeutic approach to HSN, and there remains no clear consensus on how to treat patients with severe extra-renal organ injury, progressive nephritis or nephrotic range proteinuria. Prospective randomized controlled trials seem unlikely to be viable.

The prevalence of hypertension was high at presentation (59%) and during follow-up, and was effectively treated throughout (Figure 6). This is higher than reported by others (19–42%, Table 4), although the frequency of renal impairment at presentation, also a determinant of outcome and predisposing to hypertension, was similar (32%). Hypertension at presentation and during follow-up were predictors of renal survival, the former at higher significance level. Current target pressures of 125/75 mmHg20,,21 were not achieved throughout the whole of follow-up in this series, although they were met during substantial parts (Figure 6), both in patients progressing to ESRF and in those retaining renal function. Longitudinal blood pressure data has not been described in HSN, so our findings represent the most detailed reported. Rather surprisingly, hypertension was an adverse factor only in the Italian10 study, and not in the Finnish3 or French4 series. Effectiveness of control of hypertension (Figure 6) showed little difference between the ESRF and non-ESRF groups after presentation, and may relate in importance as much to cardiovascular health as to retarding declining renal function.

Comparing our risk factors for renal failure with those of previous series, there were some similarities, but also important differences. Mean proteinuria >1 g/24 h during follow-up was a strong predictor of ESRF. In contrast, initial proteinuria >1 g/24 h was not a significant risk factor, whereas three previous studies3,4,,10 reported an adverse prognosis, although one small short-term study also found no association.16 Longitudinal data in this series on proteinuria and its significance in adult HSP suggest that (as in IgA nephropathy) modification of glomerular haemodynamics by ACE inhibitors, or AII-receptor-blockers may play a role in future treatment strategies, but there were no meaningful differences in outcome between those receiving drugs or not. The French study, with similar constraints, also found no such differences.4

Young adults in this series suffered more severe renal disease (in contrast to the French series4), but there were no major differences in clinical presentation in the two age groups. Coppo10 found that children presented with more frequent arthralgia than adults, the difference persisting (though diminishing) into adulthood. Adults in our population were younger, so the findings are not explained by their trend. The conspicuously worse outcome in males vs. females (RR 6.0; p = 0.05), 30% of males retaining independent renal function at 20 years vs. almost 90% of females, also contrasts with the findings of other studies.3,,30 This trend was exaggerated in our data, with younger males being the most seriously affected: 90% of males aged <30 years developed ESRF, and with persistent moderate proteinuria (⩾1g/24 h), all progressed.

The significance of histological findings at presentation showed some similarities with published data. Tubular atrophy and interstitial fibrosis were strong determinants of an adverse prognosis, but focal glomerulosclerosis, mesangial cell proliferation, matrix increase and arteriolar hyalinosis did not reach statistical significance (Figure 9). Much as in other vasculitides, crescents (found in one third of cases) were more frequent in patients progressing to renal failure; however, the high proportion of those retaining normal renal function with crescents (26%) was unexpected. In the French series, crescents were seen in only 4% of cases, and their presence was unrelated to outcome.4 Adult HSN thus appears similar in its outcome risks to other small-vessel vasculitic glomerulopathies, but these findings also accord with one report of crescentic disease in IgA nephropathy.31 In this report, biopsies during the active infection-induced macroscopic haematuria (comparable with acute presentation of HSN) found a high frequency of crescents, presumably transient, bearing little relationship to outcome.31 Since patients with crescent formation in acute adult HSN clearly can retain normal renal function, its natural history may be similar to that of IgA disease, with crescents not necessarily responsible for all long-term glomerular loss.32 Thus crescent formation per se in HSN should probably not be an indication for immunosuppressive treatment, unless there is continued progression; caution may be appropriate for treatment trials in crescentic disease.33,,34

Renal outcome in this series resembles the more severe forms of IgA nephropathy: the minority who progress to ESRF in <10 years.32 HSP and IgA nephropathy have been proposed to represent a spectrum of immunopathogenetically similar disorders,35 much as Churg Strauss vasculitis evolves from late-onset asthma, merging into the systemic disease. Although there are similarities in prognostic indicators between HSN and IgA nephropathy, they are quantitatively distinct, the broad spectrum of IgA nephropathy having a more benign renal course over two decades or more than in this or other series. There is a marked similarity between ESRF (27%) in this series and in other small-vessel systemic vasculitides with nephritis. Earlier 10-year figures of 35% renal loss36 in Wegener's granulomatosis were little changed more recently in all ANCA-positive renal vasculitides at 28%.37 Mortality in ANCA-positive diseases is high, one-third at 5 years being typical, but these are predominantly diseases of the elderly (median 66 years37). Mortalities in this and other series of adult (albeit younger) HSP patients, although significant, are less than half that for typical ANCA-positive renal vasculitides, doubtless a result of lesser extra-renal organ injury.

The CHCC7 criteria for Henoch Schönlein purpura, in not specifying a deposition site of immunoglobulin A, and with unreliability in reproducing skin biopsy positivity,9 mean that renal biopsy may be indicated in the absence of major urinary abnormalities. This is likely if the vasculitis type is in doubt and treatment is being considered, for example with potential gastrointestinal involvement. This UK series was not intended to be demographic, and like other published series, may thus not represent the entire spectrum of adult HSP, with under-representation of patients without urinary abnormalities. From the network and hub referral system established with dermatology clinics (the main non-acute source), with urine dipstix screening for patients presenting with cutaneous vasculitis, capture of cases is likely to be comprehensive. For acute medicine referrals, logically the more severe cases, routine positive urinalyses prompted referral. An adult with milder HSP, without urine abnormality, and with dominant cutaneous vasculitis, while possibly omitted, could not have been diagnosed with HSN, because unequivocal diagnosis requires renal biopsy. Such a patient could not reliably be diagnosed as HSP (without nephritis), since specificity and sensitivity without the renal biopsy, although high, is <90%,8 excluding such patients from descriptive series. With this caveat, the strict diagnostic criteria for Henoch Schönlein purpura with nephritis used in this report, and increasingly in others, do permit fair description and comparison of disease severity and adversity of prognosis in this UK population.


View Abstract