3.1 The key clinical outcomes for Spectra Optia presented in the decision problem were:
percentage of sickle haemoglobin (HbS)
frequency and length of procedure
staff time and group/grade needed to perform exchange transfusion
haematocrit, iron overload and need for chelation therapy
length of hospital stay for complications
venous access success rates and device‑related adverse events.
3.2 The company carried out 2 separate literature reviews, identifying a total of 33 studies including 4 that only related to adverse events; 30 of these were presented in the company submission. Only 6 of the 30 studies directly compared the Spectra Optia system, or its predecessor the Cobe Spectra system, with manual red blood cell exchange.
3.3 The external assessment centre carried out an additional literature search which identified 31 studies relevant to the decision problem, including 27 that were also identified by the company. It excluded 5 of the 30 studies presented by the company but identified 4 additional studies. Of these studies, the external assessment centre selected 13 for full evaluation. Six were chosen because they compared automated exchange with manual exchange, but they were generally of poor methodological quality (with 3 studies reported as unpublished conference abstracts). An additional 7 single‑arm studies that had been published in full in peer‑reviewed journals were included: 1 that investigated Spectra Optia and 6 that investigated Cobe Spectra.
3.4 Cabibbo et al. (2005) reported on a peer‑reviewed retrospective observational study in 20 patients with sickle cell disease who had manual or automated red blood cell exchange. In total, the authors reported 206 automated exchange procedures in 13 patients – around 30% (60/206) of which used the Cobe Spectra system and the rest used 1 of 2 other automated systems – and 188 manual exchange procedures in 7 patients. The results reported procedure time, red blood cell (RBC) units used, clinical improvement, iron overload and haemoglobin level of lower than 30% (HbS<30%) achieved, but it was not possible to compare these outcomes with baseline results. The authors concluded that the need for chelation therapy was reduced with automated exchange but that alloimmunisation increased. No statistical analysis comparing automated and manual exchange results was reported.
3.5 Dedeken et al. (2014) reported on a retrospective observational cohort study that was published as a conference abstract. In this study, 10 children (median age 11.8 years) who were having manual exchange (median 1.9 years duration) were switched to automated exchange (Spectra Optia, median 1.7 years). Results were reported separately for Spectra Optia use in years 1 and 2. Median HbS for Spectra Optia was 40% (range 28.5% to 42%) in year 1 and 46% (range 31% to 48%) in year 2 compared with 33.5% across both years (range 25% to 42%) for manual exchange (p=0.0002). The median length of procedure for Spectra Optia was 87.3 minutes and 91.0 minutes in years 1 and 2 respectively, compared with 245 minutes for manual exchange (p=0.0002). The average interval between procedures for Spectra Optia was 34 days and 42 days for year 1 and year 2 respectively compared with 28 days for manual exchange (p=0.0002). Spectra Optia used 32.2 ml/kg and 30.0 ml/kg body weight of packed RBC in year 1 and year 2 respectively, compared with 18.3 ml/kg used in manual exchange (p<0.0001). In terms of total RBC units used, Spectra Optia used 67.0 and 65.5 in year 1 and year 2 respectively, compared with 39.5 used in manual exchange (p<0.0001).
3.6 Duclos et al. (2013) reported on a retrospective case‑matched study that was published as a full article in a peer‑reviewed journal. In the study, 5 children (average age 12 years) from different treating centres had exchange with the Cobe Spectra system (60 procedures). These were matched, through weight and age, with children (average age 11 years) from a different centre who had manual exchange (124 procedures). The authors reported baseline patient data before the procedure, but post‑procedural data were not available. The transfused blood volume for treatment with Cobe Spectra was higher than that with manual exchange, at 41 ml/kg (95% confidence interval [CI] 19.6 to 60.0) compared with 11.1 ml/kg (95% CI 6.6 to 20.0).
3.7 Fasano et al. (2015) reported on a retrospective observational study that was published as a conference abstract and was not peer reviewed. The study aimed to compare the efficacy of different procedures in reducing ferritin and liver iron content. Three procedures were used: simple transfusion (top‑up transfusion, 20 patients), partial transfusion (details of procedure not reported, 6 patients) and automated exchange (system presumed Spectra Optia as stated by company, 10 patients). To be eligible, the patients needed to have a minimum of 6 months' haematological data, but these data were not reported in the abstract. Changes in ferritin and iron content were reported as well as average HbS and alloimmunisation rates. For automated exchange, the average HbS was 36% with an average ferritin change of −61 ng/ml/month (range −161 to 17). For partial transfusion, the average HbS was 34%, with an average ferritin change of 19 ng/ml/month (range −42 to 106).
3.8 Kuo et al. (2015) reported in a letter on a retrospective cohort study that was the only comparative study conducted in the UK, in 2 London centres. The aim of the study was to investigate 'whether adult sickle cell disease patients on manual exchange differ from those on automated exchange in their ability to achieve pre‑defined haematological targets, rate of complications, blood usage and clinical outcomes over a 1‑year period'. The study investigated 1 group (n=30) who had Spectra Optia for chronic sickle cell disease in 1 centre, and another group (n=21) who had manual exchange in another. The patients at each centre were not matched but were well described with no differences reported in demographics, primary indications or chelation status. However, patients having manual exchange were significantly younger than those having automated exchange with Spectra Optia (median 23 years compared with 31 years, p=0.035), and manual exchange was administered more frequently through a peripheral venous route rather than a central route (p<0.0001). The outcomes reported in the study included:
mean pre‑procedure HbS: 50% (95% CI 27% to 76%) Spectra Optia compared with 55% (95% CI 16% to 72%) for manual exchange (p=0.162)
number of patients who had less than two‑thirds of procedures within the HbS target: 19/30 Spectra Optia and 19/21 for manual exchange (p=0.048)
median post‑procedure haematocrit: 0.31 (0.23 to 0.35) for Spectra Optia and 0.31 (0.25 to 0.38) for manual exchange (p=0.931).
Resource use was also measured: average packed RBC utilisation was 55 units per patient per year for Spectra Optia and 31 for manual exchange. Procedure time was 127 minutes for Spectra Optia and 241 for manual exchange, and mean procedure intervals were 6.66 weeks for Spectra Optia and 4.86 weeks for manual exchange. Peripheral venous access was only achieved in 1 of the 30 patients in the Spectra Optia arm, whereas it was achieved in 14 of 21 patients in the manual exchange arm. Top‑up transfusions were needed in 11 procedures in the manual exchange arm, but in no patients in the Spectra Optia arm.
3.9 Woods et al. (2014) reported on a retrospective observational study that was published as a conference abstract and was not peer reviewed. In this study data were collected from 38 patients in a single institution over 2 years. The number of procedures was not reported, but in the first year 5 patients had automated exchange (confirmed to be with Spectra Optia by the company), 17 had manual exchange and 16 had both. In the second year, 13 had automated red blood cell exchange and 25 had manual exchange, but results for this year were not presented separately. Patients were actively selected for Spectra Optia based on age and size, and could choose not to have Spectra Optia. Outcomes reported in the study included: proportion of procedures achieving HbS targets (0.80 [95% CI 0.40 to 1.00]) for automated exchange and 0.50 [interquartile range 0.28 to 0.90] for manual exchange, p=0.27; ferritin concentrations (875 ng/ml [interquartile range 578–2659 ng/ml] for automated exchange and 1527 ng/ml [interquartile range 731–568 ng/ml] for manual exchange, p=0.56; and catheter complications (seen in 15 of 21 patients having automated exchange and in 1 of 17 having manual exchange).
3.10 Quirolo et al. (2015) reported on a prospective multicentre study that was published in a peer‑reviewed journal. The external assessment centre highlighted this study because it investigated Spectra Optia and was of a relatively high methodological and reporting standard. Patients (adults and children over 12 years age) were enrolled to have either standard exchange or automated exchange/deletion exchange with Spectra Optia. 72 patients were enrolled in the study, 60 of whom were evaluated for efficacy. Only 1 procedure was reported per patient. The pre‑specified primary end point was Spectra Optia's ability to accurately achieve targets for the fraction of a patient's original red cells remaining, which was reported as 0.90±0.17 (acceptable range defined as 0.75–1.25) in the evaluable population. The mean procedure time (and standard deviation) for the evaluable population was 90±22 minutes; for standard automated exchange this was 92±24 minutes, compared with 86±16 minutes for depletion exchange. The procedure time was statistically significantly longer in adults (95±24 minutes) than in children (81±16 minutes). The mean volume of replacement blood used in all procedure types was 1895±670 ml; this was also statistically significantly higher for adults (2118±702 ml) than for children (1449±260 ml). Depletion exchange needed less blood (1562±281 ml) than standard exchange (2016±729 ml), and this difference was statistically significant (although did not result in fewer units used).
3.11 Bavle et al. (2014) reported on a retrospective analysis that was published as a full article in a peer‑reviewed journal. The study analysed the physical growth of children with sickle cell disease (a secondary outcome in the decision problem) who had regular exchange. The study compared the height, weight and BMI of 36 patients before and after long‑term exchange with 2 control groups: all patients with sickle cell disease were from the Cooperative Study of Sickle Cell Disease (CSSCD), and a subset of 64 matched controls taken from CSSCD. The patients showed a significant increase in height, weight and BMI after long‑term exchange (p≤0.0001). There was also a significant increase in weight, height and BMI compared with the matched controls from the CSSCD and the entire CSSCD cohort (p<0.01).
3.12 Billard et al. (2013) reported on a retrospective case series that was published as a full article in a peer‑reviewed journal. All patients had automated exchange using the Cobe Spectra system. The study comprised 18 children having 443 procedures through a short‑term, indwelling, double‑lumen catheter, with a follow‑up of 6.5 years. Due to the descriptive nature of this case series, results were described on an individual patient basis only using a before‑and‑after analysis (Wilcoxon signed rank test), which was subject to confounding and bias.
3.13 Kalff et al. (2010) reported on a retrospective case series that was published as a full article in a peer‑reviewed journal. All patients had automated exchange in the same centre using the Cobe Spectra system. The study included 13 adult patients and evaluated the effectiveness of a regular exchange programme. Patients had red blood cell exchange through a peripheral venous cannula or arterio‑venous fistula, initially every 4 weeks and then every 4 to 6 weeks. End points included pre‑ and post‑procedure HbS (mean pre‑procedure 47.4% [range 40.7% to 59.3%], mean post‑procedure 25.5% [range 18.5% to 32.6%]), incidence of sickle cell‑related acute events, and the progression of pre‑existing related end‑organ damage and development of new end‑organ damage. The regular exchange programme reduced HbS levels to the target of <30% immediately after the procedure in all but 2 patients. A total of 16 acute sickle‑related events occurred in 5 patients in 846 cumulative months of patient follow‑up. No patient experienced stroke or multi‑organ crises, evidence of new end‑organ damage or progression of pre‑existing related end‑organ damage. Ferritin levels were monitored in 11 patients. In patients with normal baseline levels, these were maintained whereas in patients with slightly higher baseline levels they were reduced without chelation therapy.
3.14 Masera et al. (2007) reported on a retrospective review that was reported as a full article in a peer‑reviewed journal. This was an 11‑year review of routine data from a cohort of 34 patients with sickle cell disease in 1 hospital. The authors focused on 13 high‑risk patients and reported efficacy, safety and cost outcomes of a periodic regimen of erythro‑exchange with the Cobe Spectra. Outcomes included change in HbS and ferritin levels, hospital admissions and painful crises. The authors reported a reduction in all of these outcomes compared with data before erythro‑exchange was started, but the reported changes were not tested for statistical significance.
3.15 Sarode et al. (2011) reported on a retrospective review that was published as a full article in a peer‑reviewed journal. This study is a review of a 2‑phase automated exchange method using isovolaemic haemodilution with conventional red blood cell exchange (C‑RBCX), compared with the C‑RBCX protocol alone. In the study, 14 patients having the automated exchange protocol (using the Cobe Spectra device) were compared with 6 historical controls having C‑RBCX, and outcomes focused on resource use. The authors reported an increase in haematocrit (pre‑procedure 27.8±2.4%, post‑procedure 32.8±1.6%) and a decrease in HbS (pre‑procedure 41.8±6.1%, post‑procedure 9.8±2.4%) following the automated exchange protocol; the changes were not tested for statistical significance. C‑RBCX procedures needed 39.5±4.6 ml/kg packed RBC, lasted 107.3±6.7 minutes and were done every 37±7.0 days, leading to 7 procedures per year.
3.16 Shrestha et al. (2015) reported on a retrospective observational cohort study that was published as a full article in a peer‑reviewed journal. The study was designed to compare 2 methods of vascular access (dual lumen port valves with temporary central venous and peripheral catheters) during automated exchange with the Cobe Spectra system. They reported outcomes including inlet speed, duration of procedures and rates of complications. Twenty‑nine adults with sickle cell disease who had a total of 318 procedures were included for analysis. The authors reported a mean duration of 2.0±1.6 hours for the procedure and a mean number of blood units used of 6.3. They also reported 87% and 95% success rates for the post‑procedure haematocrit and HbS targets respectively.
3.17 The committee considered that there were limitations in the clinical evidence, meaning that not all of the outcomes defined in the scope could be evaluated. However, it was advised that this was partly because of limitations in study methodologies, mainly as a result of a lack of clinical equipoise in treatment modalities and the need for personalised treatment in individual patients with sickle cell disease. Nonetheless, the committee concluded that the evidence together with expert advice was sufficient to accept that Spectra Optia offers significant clinical benefits compared with manual exchange or top‑up transfusion. The committee noted that automated exchange procedures are shorter and have a longer‑lasting clinical benefit than manual exchange, meaning that patients need the procedure less often.
3.18 The committee noted that the evidence did not show significant differences in reducing iron overload in patients having automated exchange with Spectra Optia compared with patients having manual exchange. It was advised by the clinical experts that exchange transfusion was the best treatment option for avoiding iron loading in people with sickle cell disease. The committee concluded that long‑term data should be collected on how automated exchange affects iron overload status and the need for chelation therapy.
3.19 The committee was informed by the clinical experts that, in practice, manual red blood cell exchange is not iron neutral. It was advised that the level of precision needed to achieve absolute iron neutrality is not possible in a typical hospital setting or within a reasonable procedure time using the manual technique. However, the experts advised that use of Spectra Optia automated red blood cell exchange can achieve levels of precision that mean that iron neutrality can be maintained.
3.20 The committee noted that optimal iron management is very important in people with sickle cell disease. Iron chelation therapy can be used to avoid the serious complications of iron overload, but the committee was advised that this treatment is poorly tolerated and compliance is therefore low. Oral iron chelators are unpalatable and chalky and infusion pump chelators must be administered overnight and on a frequent basis. The committee accepted expert advice that Spectra Optia is the only reliably iron‑neutral transfusion therapy currently available, and that this is particularly important as chelation therapy is costly and poorly tolerated.
3.21 The committee was advised that venous access can be difficult for patients having exchange procedures, particularly in very young children (the clinical experts informed the committee that most children having Spectra Optia were over 10 years old). Experts advised that safely achieving vascular access is an important factor in adopting Spectra Optia and that this may depend on the availability of appropriately trained staff. A patient expert added that vascular access was a source of considerable anxiety for some patients before transfusion sessions. The committee was advised that there is an inequity of access to specialised venous access teams and that this may affect uptake of the Spectra Optia system. The skills needed to use Spectra Optia are transferable, so staff are able to use the device for other clinical indications.