SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Objective

To investigate the safety of therapeutic tumor necrosis factor α (TNFα) blockade in patients with systemic lupus erythematosus (SLE), in whom this proinflammatory cytokine is significantly increased and may be involved in the disease pathogenesis.

Methods

In an open-label study, 6 patients with moderately active SLE (4 with nephritis and 3 with arthritis refractory to other therapies) were given 4 300-mg doses of infliximab, a chimeric anti-TNFα antibody, in addition to immunosuppression with azathioprine or methotrexate.

Results

The only significant adverse events observed were urinary tract infection in 3 patients, 1 of which was accompanied by Escherichia coli bacteremia, and a prolonged febrile episode of putatively viral origin in 1 of them. These patients had similar infectious conditions in the past. In none of the patients was it necessary to terminate the treatment prematurely. Levels of antibodies to double-stranded DNA and cardiolipin increased in 4 patients each, but this was not associated with a decrease in serum complement levels, with vascular events, or with flares. In contrast, disease activity declined during therapy. All 3 patients with joint involvement experienced remission of arthritis, which relapsed 8–11 weeks after the last infliximab infusion. In the 4 patients with lupus nephritis, proteinuria decreased significantly within 1 week after initiation of therapy and was diminished by ≥60% within 8 weeks, remaining at low levels until the end of the observation period (at least several months).

Conclusion

Infliximab did not lead to adverse events related to an increase in SLE activity, although autoantibodies to double-stranded DNA and cardiolipin increased, as expected. This finding, coupled with the clinical improvement in the inflammatory manifestations of the disease, indicates that further study in larger controlled trials is warranted.

Systemic lupus erythematosus (SLE) is an autoimmune disease characterized by a wide variety of autoantibodies (1–5), some of which are pathogenic. The disease is often chronic and incurable, and it is associated with substantial mortality. Given that there have been no major therapeutic advances over more than 2 decades and that the current nonspecific immunosuppressive measures are linked with significant adverse events and disability, there is a need to find new therapies. Many of the novel approaches are targeted toward particular molecules or disease mechanisms (6).

The pathologic events in SLE are mediated by the formation of immune complexes (7–9), activation of the complement cascade, and engagement of cellular Fc and complement receptors (10). The ensuing inflammatory process may then lead to progressive tissue destruction and organ damage, such as lupus nephritis, a common and serious manifestation of SLE (11–13). While interference with both complement activation and with Fcγ receptor function have been shown to ameliorate experimental SLE (14–16), the role of proinflammatory cytokines in the pathogenesis of SLE as well as the effect of cytokine blockade are still unknown.

Tumor necrosis factor (TNF) is an important proinflammatory cytokine with pleiotropic properties (17–19), including the activation of a cascade of inflammatory events that lead to tissue destruction (20–22). In most studies, TNF is found to be markedly increased and appears to be bioactive in the sera of patients with active SLE, and levels of TNF have been shown to correlate with SLE disease activity (23–29). Moreover, TNF is present in renal tissue in all types of lupus glomerulonephritis and is associated with renal disease activity (30–33). These data and the finding that immune complexes induce TNF (34) indicate that blockade of TNF might have beneficial effects on organ inflammation in SLE.

However, in rheumatoid arthritis and Crohn's disease, TNF blockade leads to the formation of antinuclear autoantibodies (ANAs) in 30–40% of patients and to the formation of autoantibodies to double-stranded DNA (anti-dsDNA) in ∼15% of patients, which is occasionally associated with a transient drug-induced lupus–like syndrome (35–38). Anticardiolipin antibodies (aCL) have also been observed during TNF-blocking therapy (39). Even if many patients with SLE may be in need of new therapies because of toxicity or insufficient efficacy of their current treatments, blockade of TNF has not hitherto been considered to be an option because of the above observations as well as the controversy stemming from experimental models of the disease.

In some experimental models, TNF has been shown to have beneficial effects. The offspring of NZB/NZW mice were found to express low levels of TNF (40), and application of high doses of TNF was shown to delay disease onset in this lupus-prone mouse strain (40). There have also been recent suggestions of an important role of TNF in the down-regulation of the immune response (41). However, it was found that prolonged administration of TNF does not prevent the occurrence of nephritis in NZB/NZW mice (42). In addition, low-dose TNF administered late in the disease course was shown to cause a deterioration of nephritis (43, 44). Moreover, the kidneys of MRL/lpr lupus-prone mice contain high levels of TNF (45, 46), and serum levels of TNF in these mice correlate with disease activity (47), similar to the findings in humans (26).

Taken together, the data suggest that anti-TNF therapy may be beneficial in interfering with the organ inflammation in patients with SLE, although its potential to induce autoantibodies and to inhibit other possibly beneficial immunologic effects of TNF might constitute a significant safety issue. Before it would be ethically acceptable to start a controlled trial to evaluate the benefit of such therapy, it appears necessary to test the safety of TNF blockade in SLE patients in an open-label approach. This was the focus of the present investigation in 6 patients with SLE.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Treatment protocol.

In the conduct of this open-label safety trial, a total of 4 infusions of a fixed dose of 300 mg of infliximab (∼5 mg/kg of body weight) were scheduled. Infliximab was administered in 250 ml of saline on day 0 and in weeks 2, 6, and 10. Patients were followed up for a total of 52 weeks. The following features were monitored and analyzed: premature treatment terminations, possible adverse events (using standard clinical and laboratory parameters), changes in global disease activity as measured by the Systemic Lupus Erythematosus Disease Activity Index (SLEDAI) and the Systemic Lupus Erythematosus Index Score (SIS) (48), changes in levels of anti-dsDNA autoantibodies, C3c, C4, CH50, proteinuria (in those with nephritis), and changes in the number of swollen joints (in those with arthritis).

All patients underwent testing for tuberculosis with skin tests and chest radiographs, and no current urinary tract infections were identified on urine cultures. Baseline dosages of immunosuppressive agents (azathioprine, methotrexate, chloroquine, or prednisone) were stable throughout the treatment period.

In accordance with the Declaration of Helsinki, all 6 patients gave written informed consent to participate in this study. The informed consent form as well as the whole trial protocol were approved by the Ethics Committee of the Medical University of Vienna.

Patients.

Patients who fulfilled the American College of Rheumatology (ACR) criteria for SLE (49) were eligible to enter the study if they had low-to-moderate disease activity (defined by the SLEDAI and SIS) for at least 3 months prior to inclusion. In those with renal disease, the condition had been unresponsive to traditional cytotoxic or immunosuppressive therapy, or there were contraindications for cyclophosphamide therapy. Patients with highly active disease were excluded from the study, since no experience with the use of TNF blockade in SLE has previously been reported. For this reason, the trial was also not blinded, but each patient's clinical and laboratory features for at least 12 months preceding TNF-blocker therapy served as that patient's control.

Of the 6 consecutive patients with low-to-moderate overall disease activity who were receiving stable dosages of immunosuppressive agents and were therefore eligible for study, 5 consented to the trial and underwent treatment with the chimeric monoclonal anti-TNF antibody infliximab. An additional patient with severe nephrotic syndrome that was refractory to therapy (patient 2) was treated according to the same protocol. Even though she was not taking a stable dosage of immunosuppressive agents, she was started on azathioprine together with the infliximab regimen. This patient had a persistent nephrotic syndrome despite cytotoxic therapy and 15 cycles of immunoadsorption (50) (total processed plasma volume 108 liters), and she had a low serum albumin level (2.56 gm/dl) and severe pitting edema for many months before and at the initiation of infliximab therapy. She had received cyclophosphamide therapy for a previous episode of lupus glomerulonephritis, had undergone curative surgery for early-stage bladder carcinoma linked to the cyclophosphamide exposure, and had developed anaphylactoid reactions to treatment with 2-mercaptoethanesulfonate sodium salt, even when instilled locally into the bladder rather than systemically. Since the study protocol in this patient otherwise completely matched that in the other 5 patients, her data have been included in this report.

Four of the patients had lupus glomerulonephritis (World Health Organization class III in 1, class IV in 2, and class V in 1), and 3 had severe polyarticular lupus arthritis (1 of whom also had nephritis) (Table 1). Five patients were positive for antibodies against dsDNA. Patients 2 and 6 were also positive for anti-Ro/SSA antibodies. Among those with arthritis, patient 6 had a rheumatoid factor (RF) value of 15.1 IU/ml, just above the upper limit of the normal range (0–15 IU/ml), and patients 3 and 4 were RF-negative. All patients with nephritis (patients 1, 2, 4, and 5) had previously received cytotoxic therapy, which initially had produced significant improvement in the disease, but at study screening, they had a recurrence of proteinuria despite therapy with immunosuppressive agents and prednisone. Patient 2, who had nephrotic syndrome as described above, also had an active urinary sediment, with erythrocyte and granular casts. Patient 5 likewise had nephrotic-range proteinuria despite attempts to treat the condition with cyclosporin A and azathioprine. Patients 4 and 5 had microhematuria, and patient 1 had sterile leukocyturia in urinary sediment at baseline.

Table 1. Characteristics of the SLE patients at the screening visit*
CharacteristicPatient 1Patient 2Patient 3Patient 4Patient 5Patient 6
  • *

    SLE = systemic lupus erythematosus; ACR = American College of Rheumatology; WHO = World Health Organization; GN = glomerulonephritis; CSA = cyclosporin A; IAS = immunoadsorption; LEF = leflunomide; AZA = azathioprine; CYC = cyclophosphamide; ANA = antinuclear antibody; anti-dsDNA = anti–double-stranded DNA; SLEDAI = Systemic Lupus Erythematosus Disease Activity Index; SIS = Systemic Lupus Erythematosus Index Score.

  • In patient 2, AZA was initiated concomitantly with infliximab, whereas in all other patients, baseline therapy had been stable for at least 3 months before initiation of infliximab.

Sex/age, yearsF/26F/52F/57F/54F/25F/32
SLE duration, years1729920810
No. of ACR criteria fulfilled565764
Current dominant organ involvementKidneyKidneyJointsKidney; jointsKidneyJoints
WHO class GN on biopsyIIIIVIVV
Baseline treatment      
 Prednisone10 mg/day25 mg/day5 mg/day6.25 mg/day10 mg/day10 mg/day
 Azathioprine50 mg/day100 mg/day100 mg/day50 mg/day125 mg/day
 Methotrexate20 mg/week
 Chloroquine250 mg/day250 mg/day250 mg/day
Previous treatment      
 Immunomodulatory drugsCSAIASLEFAZACSA
 CYC, total dose9 gm14 gm19 gm3 gm
 Time since last CYC bolus>1 year4 months>1 year>1 year
Proteinuria, gm/24 hours1.25.7<0.053.16.4<0.05
ANA titer (normal <1:80)1:1601:1,2801:2,5601:6401:6401:640
Anti-dsDNA, IU/ml (normal 0–7)31.024.496.950.75.114.2
Other ANA subset antibodiesAnti-RoAnti-Ro
Complement      
 C3c, mg/dl (normal 90–180)84.220.992.589.862.472.6
 C4, mg/dl (normal 10–40)18.3<6.710.57.710.47.5
 CH50, % (normal 70–140)127<1014013583.8140
SLEDAI6861479
SIS5871186

Patients 3, 4, and 6 had persistent polyarthritis despite treatment with 20 mg of methotrexate per week (patient 3) or ≥100 mg of azathioprine per day (patients 4 and 6). In order to provide basic immunosuppression during concomitant administration of infliximab, patient 3 continued taking methotrexate, patients 1, 4, 5, and 6 continued taking azathioprine (as they had during many months before infliximab), and patient 2 began daily therapy with azathioprine (100 mg/day) when infliximab was initiated.

Determination of autoantibodies.

ANAs were detected by indirect immunofluorescence on Hep-2 cells (Aventis Pasteur, Marcy L'Etoile, France). Anti-dsDNA antibodies were detected by radioimmunoassay (Ortho Clinical Diagnostics, Rochester, NY). Antibodies against Ro/SSA, La/SSB, Sm, and U1 RNP (and other ANA subsets) were detected by the line immunoassay (Innogenetics, Ghent, Belgium) and by immunoblotting using HeLa cell extracts as described elsewhere (51, 52). In addition, IgG and IgM aCL autoantibodies, lupus anticoagulant (LAC), and RF were determined.

RESULTS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Adverse events and premature treatment termination.

None of the 6 patients had to terminate the study prematurely, and no infusion reactions were observed. Likewise, no increase in disease activity was observed (see below).

After the second infusion of infliximab, 1 patient (patient 4) had a febrile episode, with subsequent thrombocytopenia and macrohematuria (from previously diagnosed benign bladder lesions, as determined by cystoscopy and by microscopy showing the absence of dysmorphic erythrocytes), which prompted her admission to the hospital and a 12-day delay in administering the third infusion of infliximab. Blood and urine cultures for bacteria and tests for cytomegalovirus infection (by serologic, early antigen [pp65] detection, and polymerase chain reaction techniques) yielded negative results. This episode was not associated with a decrease in levels of serum complement components or an increase in levels of anti-dsDNA antibodies. Moreover, this patient did not have thrombocytopenia before or after this episode, no other organ involvement was noted, and renal function remained unchanged. The episode subsided without specific therapy and did not recur when infliximab was continued. Therefore, this intercurrent event was not regarded as constituting a lupus flare, but rather, a transient thrombocytopenia in the context of an unspecified viral infection.

A few weeks later, the same patient (patient 4) had a urinary tract infection that was complicated by Escherichia coli bacteremia. This responded to treatment with oral antibiotics and did not lead to protocol changes. Patients 1 and 5 experienced uncomplicated urinary tract infections, which were also treated with short courses of oral antibiotics. All 3 patients had previously had similar urinary tract infections during the year before infliximab therapy.

Clinical effects on SLE.

None of the patients had an increase in disease activity while receiving infliximab or during a subsequent period of up to 52 weeks of observation. Indeed, within the 10-week treatment period, global disease activity decreased in all patients (Figure 1). The SIS fell from a mean ± SD 10 ± 3 to 6 ± 2 (P < 0.001 by paired t-test), the SLEDAI declined from 9 ± 3 to 5 ± 2 (P < 0.002 by paired t-test). After the fourth infliximab infusion, disease activity continued to be low in patients without arthritis (Figure 1A), but increased again in patients with arthritis (Figure 1B), and this was solely due to flares of joint disease.

thumbnail image

Figure 1. Decrease, rather than flare, in disease activity with tumor necrosis factor–blocking therapy in patients with systemic lupus erythematosus. Disease activity, as measured by the Systemic Lupus Erythematosus Index Score (SIS) and the Systemic Lupus Erythematosus Disease Activity Index (SLEDAI), is shown for patients without (A) and with (B) arthritis. Arrows indicate infliximab infusions (starting at time 0). Numbers across the bottom indicate the number of weeks after initiation of infliximab therapy.

Download figure to PowerPoint

Effects on arthritis.

In all 3 patients with arthritis, joint swelling subsided completely within a short time after the initiation of infliximab (Figure 2). A similar effect was seen for joint tenderness. In patients 3 and 6, remission was sustained until 8 weeks after the last infusion, when their arthritis relapsed. Patient 4 began to experience a relapse at 11 weeks after her last infusion (Figure 2). Out of protocol, and after several in-depth discussions with the patient and with her additional written consent, patient 6 was given an additional infusion of 300 mg of infliximab at week 20, the end of the predefined observation period (10 weeks after the last per-protocol infusion). This infusion induced remission of her arthritis flare within 24 hours. Again, the patient experienced no adverse events during this second exposure to infliximab. For the sake of completeness, these data are also shown in Figure 2.

thumbnail image

Figure 2. Decrease in swollen and tender joint counts with tumor necrosis factor–blocking therapy in patients with systemic lupus erythematosus. Shown are the numbers of swollen and tender joints in the 3 patients who had polyarthritis. Arrows indicate infliximab infusions (starting at time 0). Numbers across the bottom indicate the number of weeks after initiation of infliximab therapy.

Download figure to PowerPoint

Effects on nephritis.

In all 4 patients with nephritis, proteinuria decreased after the start of infliximab therapy (Figure 3), reaching the lowest level around week 8 and consistently remaining low thereafter. Of note, all of these patients had significant persistent proteinuria for months to years before infliximab therapy. In patient 2, who had nephrotic syndrome before infliximab therapy, pitting edema resolved within 2 weeks after initiation of infliximab, and there was an ∼30% increase (from 2.6 to 3.3 gm/dl) in the serum albumin level.

thumbnail image

Figure 3. Decrease in proteinuria and stable normal levels of serum creatinine with tumor necrosis factor–blocking therapy in patients with systemic lupus erythematosus. Shown are the course of proteinuria and the serum creatinine levels in the 4 patients who had kidney involvement. Arrows indicate infliximab infusions (starting at time 0). Numbers across the bottom indicate the number of weeks before and after initiation of infliximab therapy.

Download figure to PowerPoint

The decrease in proteinuria was not due to a reduction of glomerular filtration in any of the patients, since their serum creatinine levels remained stable and within the range of normal (Figure 3), as did their creatinine clearance rate (mean ± SD 79 ± 29 ml/minute at baseline and 100 ± 32 ml/minute at week 20). Thus, in these 4 patients, in whom proteinuria had not been sufficiently controlled with immunosuppressive therapies, TNF blockade with infliximab significantly diminished urinary protein loss by ≥60%. Moreover, no urinary casts were detectable after the third infusion of infliximab. In all patients, stabilization of nephritic disease was sustained through many weeks after the last infusion (Figure 3). Indeed, proteinuria remained at a low level for more than 6 months after the last infliximab infusion.

Effects on anti-dsDNA and complement levels.

Levels of ANAs and anti-dsDNA, as measured by radioimmunoassay, remained stable during the first few weeks of infliximab therapy. However, in patients 2, 3, 4, and 6, anti-dsDNA levels increased (from 34 to 112, from 87 to 196, from 51 to 193, and from 14 to 72 IU/ml, respectively) around the time of the fourth infliximab infusion (Figure 4). Anti-dsDNA levels tended to stabilize or decrease with cessation of anti-TNF therapy, but further increased in patient 6 after the reinstitution of infliximab.

thumbnail image

Figure 4. Stable complement levels but a transient increase in anti–double-stranded DNA (anti-dsDNA) antibody levels with tumor necrosis factor–blocking therapy in patients with systemic lupus erythematosus. Shown are the levels of anti-dsDNA antibodies (measured by radioimmunoassay) and serum C3c in all 6 patients. Arrows indicate infliximab infusions (starting at time 0). Numbers across the bottom indicate the number of weeks before and after initiation of infliximab therapy.

Download figure to PowerPoint

As mentioned above, no new clinical manifestations or clinical deterioration beyond the recurrence of arthritis (joint counts returning to baseline) were seen. In fact, although anti-dsDNA antibodies are one of the variables of both the SLE activity scores we used, no significant increase in disease activity was observed (Figure 1). This was also consistent with the stable levels of C3c that were observed (Figure 4).

Four patients (patients 2, 4, 5, and 6) experienced a transient increase in IgM aCL autoantibodies (from 11.3 to 21.7, from 4.2 to 19.5, from 10.1 to 47.1, and from 3.4 to 33.4 IgM phospholipid units per ml). Patient 2 also had a transient increase in IgG aCL (from 7.3 to 25.1 IgG phospholipid units per ml) with a subsequent decline (to 3.8 IgG phospholipid units per ml). Two patients who were positive for LAC at baseline (patients 5 and 6) tested positive throughout the trial. The other 4 patients remained negative for LAC. No vascular events were observed during or after infliximab therapy.

DISCUSSION

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

TNF blockade as therapy for SLE is a highly controversial issue. The main concerns derive from some data in experimental models of the disease (53) and from the induction of ANAs, anti-dsDNA, and aCL antibodies and some rare cases of drug-induced lupus–like syndromes in patients treated with anti-TNF agents (35–39). Indeed, among the 6 SLE patients we treated with infliximab, 4 had an increase in anti-dsDNA antibodies. In addition, aCL antibodies increased in 3 of these 4 patients and in an additional patient. This increase in both anti-dsDNA and aCL antibodies during anti-TNF therapy suggests that TNF, in fact, may down-modulate autoantibody responses.

However, neither clinical deterioration of the underlying disease nor a drug-induced lupus–like syndrome was observed. One could argue that the observed relapses of arthritis in the 3 patients with polyarthritis were signs of lupus flares. However, SLE disease activity did not increase to levels higher than they were before therapy, there was no new organ involvement, renal disease remained improved in the patient with concurrent arthritis and nephritis, and most important, rechallenge with another infusion of infliximab in patient 6 had the same beneficial clinical effect. Complement levels did not decrease, but rather, improved in some of the patients, further supporting the notion that SLE activity did not increase following infliximab. As with previous experience indicating that autoantibodies (and even lupus-like episodes) observed in rheumatoid arthritis and Crohn's disease patients usually resolve with cessation of anti-TNF treatment, the anti-dsDNA antibodies did not further increase or decrease during the followup observation period after the fourth (and last) infusion.

None of the patients had to terminate the study prematurely, and none had an infusion reaction. However, 3 patients experienced a urinary tract infection, and the infection was complicated by E coli bacteremia in 1 of them. Although all 3 patients had a history of recurrent urinary tract infections, the observation made is consistent with an increased risk of infection during treatment with anti-TNF agents, which confirms previous indications of such an effect.

This study was primarily designed as a first open trial to examine the safety of infliximab in SLE and not to prove its efficacy. However, the observed clinical findings under TNF blockade suggest that infliximab may have a therapeutic effect in patients with SLE. In particular, the reduction of proteinuria by ≥60% within a few weeks, including normalization of urinary protein in patient 1 and a striking effect even on nephrotic-range proteinuria and the consequent nephrotic syndrome in patient 2, is noteworthy. Of interest, proteinuria and hematuria, which had been present in all 6 patients for many months or years prior to this study, did not recur in any of them during an observation period of up to 52 weeks (and up to >8 months off infliximab therapy). Evidence of the potential clinical benefit of infliximab is further supported by the effects on hitherto refractory lupus arthritis, which completely remitted in all 3 affected patients, although it recurred after cessation of infliximab treatment.

The courses of the disease before infliximab therapy suggest that the observed improvement in nephritis and arthritis were not the results of mere chance or of treatment with the other therapeutic agents. In particular, except for 1 patient, immunosuppressive therapy had been stable for several months before the initiation of anti-TNF therapy. Moreover, the consistent improvement in objective laboratory measures, such as proteinuria, in all of the patients with nephritis and within a short time after the first infusion of infliximab speak for a true clinical benefit—with all caveats in mind. Further indications of clinical benefit are the stable serum creatinine levels and stable or improving serum C3c levels as well as the overall clinical improvement observed. All of these findings must be interpreted with great caution given the open nature of this study. However, small open trials have provided important clinical information that has led the way to infliximab therapy in rheumatoid arthritis, ankylosing spondylitis, and psoriatic arthritis (54–56), and similar trials in SLE have laid the groundwork for subsequent clinical trials (57).

Thus, our data allow us to speculate that anti-TNF therapy could suppress the local tissue destruction in SLE. However, at the same time, and unassociated with clinical consequences, TNF blockade may lead to increased autoantibody formation (53, 58). These 2 differential effects, taken together with the finding of a very rapid and sustained improvement in renal disease, suggest that anti-TNF agents may be, but possibly should only be, used as rescue therapy in patients with refractory forms of the disease.

In summary, although no firm conclusions can be drawn from this open-label study of TNF-blocking therapy with infliximab in SLE, 2 patterns seem to emerge. First, it appears likely that a short course of anti-TNF therapy may be effective in SLE, perhaps as a consequence of its antiinflammatory effects. Second, and not unexpectedly, there is a trend toward increased autoantibody production, although there is no indication in these patients that the enhanced autoreactivity caused clinically untoward effects, at least during an observation period of up to 52 weeks (42 weeks after the last infliximab infusion). Despite the efficacy observed, these data must be interpreted with all necessary caution, since previous uncontrolled trials in lupus have been misleading in this respect. Moreover, the present study was small and was limited to patients with nephritis and/or arthritis and low-to-moderate overall disease activity. Thus, to sufficiently address the potential value of TNF-blocking therapy in SLE, larger and controlled clinical trials are necessary. The data presented here justify such trials.

REFERENCES

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES