Evidence review

Advice

Clinical effectiveness
Safety and tolerability
Evidence strengths and limitations

Evidence review

A literature search was conducted which identified 17 references (see search strategy for full details). These references were screened using their titles and abstracts and 13 references were obtained and assessed for relevance.

Of these references, 7 were included in this evidence summary:

1 open-label randomised controlled trial (RCT) (Daoussis et al. 2010) and an associated follow-up study (Daoussis et al. 2012)
1 open-label non-randomised comparative study (Daoussis et al. 2016)
1 retrospective nested case-control study (Jordan et al. 2015)
3 prospective open-label non-comparative observational studies (Bosello et al. 2015, Lafyatis et al. 2009 and Melsens et al. 2016).

A summary of the included studies is shown in table 2 (see evidence tables for full details). Only evidence relating to skin involvement in systemic sclerosis is discussed in the evidence summary because NHS England is developing a clinical commissioning policy on Rituximab for connective tissue disease with interstitial lung disease.

Table 2 Summary of included studies

Study	Population	Intervention and comparison	Primary outcomes
Daoussis et al. (2010) Single centre prospective open-label RCT Plus Daoussis et al. (2012) Follow-up study	14 people with diffuse SSc and ILD	8 participants born on an even-numbered date received rituximab 375 mg/m² weekly for 4 weeks at baseline and at 6 months in addition to their usual treatment 6 participants born on an odd-numbered date received their usual treatment only	Changes in lung function tests and mRSS
Daoussis et al. (2016) Multicentre prospective open-label non-randomised comparative study	51 people with SSc-associated ILD	33 participants chose to receive rituximab 375 mg/m² weekly for 4 weeks at baseline, repeated 6 monthly for at least 2 cycles (with or without other immunosuppressants) 18 participants declined rituximab and received conventional treatment only	Changes in lung function tests and mRSS
Jordan et al. (2015) Multicentre, retrospective nested case-control study	63 people with SSc who had been treated with rituximab	Several dosage regimens were used. 75% of participants received 2 infusions of rituximab 1,000 mg 2 weeks apart 25 of the rituximab participants with severe diffuse disease were matched with controls from the EUSTAR database who had not been treated with rituximab	Change in mRSS
Bosello et al. (2015) Open-label, prospective non-comparative study	20 people with progressive, diffuse cutaneous SSc	2 infusions of rituximab 1,000 mg 2 weeks apart No comparator	Changes in mRSS and lung function tests
Lafyatis et al. (2009) Open-label, prospective non-comparative study	15 people with early diffuse cutaneous SSc	2 infusions of rituximab 1,000 mg 2 weeks apart No comparator	Change in mRSS
Melsens et al. (2016) Open-label, prospective non-comparative study	14 people with early diffuse cutaneous SSc	2 infusions of rituximab 1,000 mg 2 weeks apart at 0 and 6 months No comparator	Sensitivity to change of the EScSG activity index
Abbreviations: EScSG, European Scleroderma Study Group; EUSTAR, European Scleroderma Trial and Research; ILD, interstitial lung disease; mRSS, modified Rodnan skin score; RCT, randomised controlled trial; SSc, systemic sclerosis

The remaining 6 references were excluded. These are listed in excluded studies with reasons for their exclusion.

Clinical effectiveness

An overview of the study results for clinical effectiveness can be found in the results tables.

Skin thickness: modified Rodnan skin score (mRSS)

In the open-label RCT by Daoussis et al. (2010), in people with diffuse systemic sclerosis and interstitial lung disease receiving rituximab (n=8), a statistically significant improvement was seen in skin thickness (assessed using the modified Rodnan skin score [mRSS]) at 1 year compared with baseline (mean improvement 5.13, p=0.0003). The improvement was within the range considered to include the minimum clinically important value (range 3 to 7.5 [Gazi H et al. 2007]). By comparison, no statistically significant difference from baseline was seen in participants receiving usual care (n=6).

The median percentage improvement in mRSS from baseline to 1 year was 39.25% in the rituximab group compared with 20.80% in the usual care group; however, the difference between the groups did not reach statistical significance (p=0.06). Daoussis et al. (2010) noted that this might be because of the small number of participants in the study and the short period between biopsies (24 weeks).

The uncontrolled follow-up study by Daoussis et al. (2012) found that, at 2 years, mRSS had improved further in the participants who received rituximab in Daoussis et al. (2010) compared with baseline (mean improvement 8.63, which is considered clinically important, p<0.0001). The median percentage improvement in mRSS from baseline increased to 64.58% at 2 years (no statistical analysis).

In their open-label non-randomised comparative study in people with systemic sclerosis and interstitial lung disease, Daoussis et al. (2016), found statistically significant improvements from baseline in mRSS with rituximab (n=33) at all time points (1, 2, 3 and 4 years, mean improvements 5.89 to 10.19, all p<0.01, all clinically important). Statistically significant improvements from baseline were also seen with usual care (n=18) at 2, 3 and 4 years (mean improvements 2.25 to 4.14, all p<0.05), but not 1 year.

Daoussis et al. (2016) compared the 2 groups and found that rituximab was better than usual care at improving mRSS at 1, 2 and 3 years (differences 3.89, 4.88 and 7.94 respectively; p=0.002, p=0.015 and p=0.002) but not at 4 years. The differences between the groups at 1–3 years were statistically significant and within or above the range considered to include the minimum clinically important value.

In their retrospective nested case-control study in 25 people with severe, diffuse systemic sclerosis, Jordan et al. (2015) found a statistically significant improvement in mRSS from baseline in the rituximab group at 6 months (mean improvement 6.3, which is clinically important, p=0.0001). No statistically significant difference from baseline was seen in the control group. The difference in mRSS between rituximab and control was 4.4, which was statistically significant (p=0.02) and within the range considered to include the minimum clinically important value.

The nested case-control analysis also found a statistically significant improvement in the median percentage improvement in mRSS from baseline to 6 months in the rituximab group compared with the usual care group (24.0% compared with 7.7% respectively, p=0.03).

In addition to the network analysis, Jordan et al. (2015) prospectively looked at mean changes in mRSS in 46 people with systemic sclerosis and a subgroup of these who had diffuse disease (n=35). They found statistically significant improvements from baseline with rituximab in both groups, which were within the range considered to include the minimum clinically important value (mean improvements 3.7 and 4.4 respectively, p=0.0002 and p=0.0005 respectively).

Three prospective open-label non-comparative observational studies have assessed the effects of rituximab in diffuse cutaneous systemic sclerosis: Bosello et al. (2015) (n=20) in progressive disease and Lafyatis et al. (2009) (n=15) and Melsens et al. (2016) (n=14) in early disease (less than 18 months and less than 4 years respectively).

Bosello et al. (2015) found statistically significant improvements in mRSS from baseline with rituximab at all time points (all p<0.0001 except 6 months, p=0.001). The mean improvements from 6 months to 4 years ranged from 7.9 to 14.2, which are above the values considered clinically important.

Melsens et al. (2016) also found statistically significant improvements in mRSS from baseline with rituximab at all time points (all p<0.001). The score improved over time from baseline to 12 months: the mean improvement was 5.9 at 3 months and 14.4 at 12 months.

The study by Lafyatis et al. (2009), which had a 12‑month follow‑up, was the only study included in this evidence summary which found no statistically significant changes in mRSS from baseline with rituximab.

See the results tables for more details.

Functional disability: Health Assessment Questionnaire Disability Index (HAQ‑DI)

In their open-label RCT in people with diffuse systemic sclerosis and interstitial lung disease, Daoussis et al. (2010) found a statistically significant improvement in functional impairment (assessed using the HAQ-DI) at 1 year in participants receiving rituximab (n=8), compared with baseline (median improvement 0.38, p=0.03). The improvement was above the range considered to include the minimum clinically important value (range 0.2 to 0.25 [Gazi H et al. 2007]). No statistically significant difference from baseline was seen in participants receiving usual care (n=6). It is not reported whether there was any significant difference between rituximab and usual care.

In the rituximab group in Daoussis et al. (2010), functional impairment improved in 6 participants (shown by an improvement in HAQ‑DI of more than 0.2) and was unchanged in 2 participants over 1 year. By comparison, in the usual care group, 3 participants improved, 2 were unchanged and 1 participant got worse.

The uncontrolled follow-up study by Daoussis et al. (2012) found that, at 2 years, HAQ‑DI had continued to improve in the participants receiving rituximab compared with baseline (median improvement 0.44, which is considered clinically important, p<0.0001).

In the prospective open-label non-comparative study in people with diffuse cutaneous systemic sclerosis by Bosello et al. (2015) (n=20), statistically significant improvements in HAQ‑DI from baseline were seen with rituximab at all time points (all p<0.0001). The mean improvement remained consistent at 0.5 to 0.6 (clinically important) over a mean follow-up of 48.5 months.

Small improvements in HAQ‑DI were seen in people with early (less than 18 months) diffuse cutaneous systemic sclerosis in the prospective open-label non-comparative study by Lafyatis et al. (2009) (n=15); however, these did not reach the level considered to be clinically important (mean improvement 0.03 at 6 months and 0.12 at 12 months). The statistical significance of the improvements was not reported.

HAQ‑DI was not assessed in the studies by Daoussis et al. (2016), Jordan et al. (2015) or Melsens et al. (2016).

See the results tables for more details.

Disease activity and severity: European Scleroderma Study Group (EScSG) indices

In their prospective open-label non-comparative study in people with diffuse cutaneous systemic sclerosis (n=20), Bosello et al. (2015) found that statistically significant improvements in disease activity (assessed by the EScSG index) from baseline were seen with rituximab at all time points (all p<0.0001). The mean improvement remained consistent at 4.0 to 4.3 over a mean follow-up of 48.5 months. It is unclear whether this improvement is clinically important.

Statistically significant improvements from baseline were also seen in the EScSG severity index at all time points (all p<0.0001) in the study by Bosello et al. (2015). The mean severity improved over time from baseline to 4 years. The mean improvement was 3.4 at the last available follow-up; however, the clinical importance of this is uncertain.

A steady improvement in the EScSG disease activity index was seen in people with early (less than 4 years) diffuse cutaneous systemic sclerosis in the prospective open-label non-comparative study by Melsens et al. (2016) (n=14). The improvement from baseline was statistically significant at all time points (all p<0.001) and at 12 months reached 3.6. The minimum clinically important treatment effect is unclear for the EScSG indices.

EScSG disease activity and severity indices were not assessed in the studies by Daoussis et al. (2010), Daoussis et al. (2012), Daoussis et al. (2016), Jordan et al. (2015) or Lafyatis et al. (2009).

See the results tables for more details.

Safety and tolerability

Several deaths were reported in the studies included in the evidence summary. In the open-label non-randomised comparative study by Daoussis et al. (2016), 5/33 participants (15%) receiving rituximab died (3 with end-stage lung fibrosis, 1 with lung cancer, and 1 with sudden death of unknown cause) compared with 2/18 (11%) receiving usual care (both with respiratory tract infections: no statistically significant difference between the groups). Two deaths were reported in the study by Bosello et al. (2015) (1 with heart failure and 1 with arrhythmia), which were considered probably unrelated to study medicines.

Adverse events were reported in all of the studies but the severity of adverse events is not always clear. In the open-label RCT and follow-up study by Daoussis et al. (2010) and Daoussis et al. (2012), 2/8 participants in the rituximab group were hospitalised (with respiratory tract infections) and 1 experienced a mild infusion reaction, whereas 0/6 participants in the usual care group reported adverse events (no statistical analysis). In Daoussis et al. (2016), 3/33 participants (9%) receiving rituximab were hospitalised (with respiratory or urinary infections) compared with 10/18 (56%) receiving usual care (9 with respiratory or urinary tract infections and 1 with digital ulcers: no statistical analysis).

Overall, the adverse effects seen in the studies reflect those listed in the summary of product characteristics for rituximab. The most commonly reported adverse effects were respiratory tract, urinary tract and other infections (sometimes leading to hospitalisation), infusion-related reactions (generally mild), fatigue, rigors and nausea. Serious adverse events reported in 1 or 2 participants included hepatitis B virus reactivation, and cardiac and renal adverse events. An overview of the study results for safety and tolerability can be found in the results tables.

According to the summary of product characteristics, signs and symptoms suggestive of an infusion-related reaction have been reported in more than 50% of participants in clinical trials across rituximab's licensed indications, with 12% of participants experiencing severe reactions. Severe infusion-related reactions with a fatal outcome have been reported in post-marketing use. Premedication with an antipyretic and an antihistamine should be given before administration of intravenous rituximab. In addition, premedication with a glucocorticoid should be given (except in people with non-Hodgkin's lymphoma or chronic lymphocytic leukaemia who are receiving rituximab in combination with glucocorticoid-containing chemotherapy).

Very rare cases of fatal progressive multifocal leukoencephalopathy have been reported after use of rituximab and people should be monitored at regular intervals for any new or worsening neurological symptoms or signs suggestive of this condition.

Serious infections, including fatalities can occur during rituximab therapy, and rituximab is contraindicated in people with an active, severe infection (for example, tuberculosis, sepsis and opportunistic infections), and in people who are severely immunocompromised. In addition, cases of hepatitis B reactivation, including those with a fatal outcome, have been reported in people receiving rituximab. Hepatitis B virus screening should be performed in all people before starting treatment with rituximab and people with active hepatitis B infection should not be treated with this medicine.

Severe skin infections, such as toxic epidermal necrolysis and Stevens−Johnson syndrome (some with fatal outcome), have been reported in people receiving rituximab. Treatment with rituximab should be stopped if such an event occurs.

See the summary of product characteristics for rituximab for full details of warnings, contraindications and adverse events.

Evidence strengths and limitations

The studies included in the evidence summary have many limitations that affect their application to clinical practice. All included small numbers of participants and most were open-label, observational studies without a comparator (Bosello et al. 2015, Lafyatis et al. 2009 and Melsens et al. 2016). Observational studies are particularly susceptible to bias, confounding and other methodological problems, and these studies cannot prove that systemic sclerosis improved because of rituximab: they can only suggest that rituximab was associated with an improvement in the condition.

In the majority of the studies, there was no blinding of treatment or outcome assessment. Also, participants in many studies were receiving concomitant immunosuppressants and it cannot be excluded that these contributed to any improvement in symptoms. Similarly, many participants received concomitant pre-treatment with methylprednisolone. It is important to note that, many of the improvements seen in mRSS were of borderline clinical importance only. Nevertheless, results are based on the whole cohort and it is possible that some individual participants will have seen clinically important improvements.

In the studies where a comparator was used, the method of treatment allocation was not ideal and allocation concealment was not reported. The comparative study by Daoussis et al. (2016) was not randomised: participants chose whether to receive rituximab or usual care. In addition, the treatment and control groups were heterogeneous in terms of disease duration. In the study by Daoussis et al. (2010) participants were randomised according to their date of birth, which led to unbalanced numbers in each group. Also, this study included only 14 participants and was not sufficiently statistically powered to reliably detect any differences between the groups.

The nested case-control analysis by Jordan et al. (2015) has some further limitations. For example, it is a multicentre study and, although this means it reflects clinical experience in Europe, there may be differences in ratings of outcomes across investigators and centres. Some data were collected retrospectively, which could lead to recall bias.

Skin thickening often improves spontaneously during the natural course of scleroderma, making it difficult to determine whether any improvements seen with rituximab are related to the treatment or not, particularly in people with longstanding disease. Daoussis et al. (2012), noted that the median percentage improvement seen in their study (n=8) was 40% and that this is higher than the 20% expected during the natural course of the disease. For comparison, Jordan et al. (2015) found that the median percentage improvement in mRSS from baseline to 6 months in the rituximab group was 24.0% compared with 7.7% in the usual care group in their nested case-control analysis (n=25), with a statistically significant difference between the groups (p=0.03).

Participants in the comparative studies had longstanding scleroderma (average duration around 6 to 8 years) and had previously received a variety of immunosuppressants. They may have had disease that was resistant to treatment, which was more likely to deteriorate over the course of the study, or they may have been at a stage in the disease where spontaneous resolution was likely. Overall, the studies mostly included people with diffuse systemic sclerosis, but some of the study populations were heterogeneous in terms of disease duration, severity and previous treatments.

Although benefits in terms of skin thickening, functional impairment and disease activity and severity were seen in the majority of the studies, these limitations should be taken into account. The study populations varied and it is unclear which people might benefit most from treatment. Also, several different dosage regimens were used (primarily 375 mg/m² weekly for 4 weeks at baseline and 6 months, and 1,000 mg at 0 and 2 weeks); therefore, it is unclear which dosage is optimal and how often treatment cycles need to be repeated.

An overview of the quality assessment of each included study can be found in the evidence tables.