溃疡性结肠炎的系统性综述

Systematic review: how effective are the usual treatments for ulcerative colitis? J . R. BEBB & B. B. SCOTT Department of Gastroenterology, Lincoln County Hospital, Lincoln, UK Accepted for publication 3 May 2004 SUMMARY Background: Details of the efficacy of the drugs used in ulcerative colitis are not readily available. Methods: We have reviewed all placebo-controlled trials of the commonly used drugs for both induction and maintenance of remission to determine the efficacy and to calculate the numbers needed to treat (NNTs) to achieve a specified benefit for each drug. Results: The drug response rates and the NNTs (with 95% CI) are tabulated for each drug. Conclusion: Corticosteroids give a remission rate of 68% in mild or moderate disease and an NNT for remission of 2 (95% CI 1.4–5) in mild disease. Intravenous hydro- cortisone gives a remission rate of 60–73%. Aminosal- icylates are relatively ineffective in inducing remission with an NNT of 10 (95% CI 7–21) improving to 8 (95% CI 5–20) if the dose ‡3 g daily. They are better at maintenance (NNT ¼ 6; 95% CI 4–8). Intravenous ciclosporin is very effective in achieving remission in severe colitis with an NNT of 1.2 (95% CI 1–2.5). Although there is fairly good evidence that azathioprine is effective in maintaining remission and is used widely, there are no suitable placebo-controlled trials to calcu- late the NNT. INTRODUCTION In December 2003 a vice president of GlaxoSmithKline caused a furore in the media when he told a scientific meeting in London that the vast majority of drugs only work in 30–50% of people.1 That such a statement can cause a furore suggests that most people have the belief that most drugs work in most people. Indeed most clinicians hold the naı¨ve belief that the drug they are prescribing will work and they usually do not prepare themselves or their patient for the possibility of failure. Of course, this has the advantage of improving response; we all tend to respond better to a treatment if we have faith in it. However, we also need to be honest with our patients especially when they ask about the benefits of a proposed treatment. Doctors also need to be aware of likely efficacy of treatments when balancing risks and benefits. The purpose of this paper is to reveal the efficacy of drugs used in colitis. There are relatively few drug treatments that gastro- enterologists in a nonresearch environment use to induce and maintain remission of ulcerative colitis. These can be split into the following: corticosteroids (hydrocortisone and prednisolone); aminosalicylates or 5-ASA-based drugs (including sulfasalazine, mesalazine, olsalazine and balsalazide); azathioprine (AZA) and its active constituent mercaptopurine (MP) and ciclosporin. In our local practice, we routinely provide patients with information sheets on the commonly prescribed treatments, with data such as side-effects, important interactions etc. We have not hitherto informed our patients, or ourselves, of the likely benefits (over placebo or no treatment) of taking the treatments we recom- mend, either in achieving remission or maintaining it. Correspondence to: Dr B. B. Scott, Department of Gastroenterology, Lincoln County Hospital, Lincoln, LN2 5QY, UK. E-mail: drbbscott@aol.com Aliment Pharmacol Ther 2004; 20: 143–149. doi: 10.1111/j.1365-2036.2004.02018.x � 2004 Blackwell Publishing Ltd 143 To correct this omission we have reviewed all placebo- controlled trials of drugs commonly used in ulcerative colitis and calculated an expected remission rate (or maintenance of remission rate) for placebo and the particular drug. Based on these data we have calculated the number needed to treat (NNT) for each treatment to induce or maintain remission. Where insufficient pla- cebo-controlled data exist, we have given a ‘best guess’ as to the expected remission rate with the drug, based on nonplacebo-controlled data. METHODS A computer-assisted search using the OVID interface to MEDLINE was conducted to identify potentially relevant published papers. A search of theMEDLINE database from 1966 to 2003 was performed using the terms ‘ulcerative colitis’, ‘placebo’ and the names of the following drugs: ‘corticosteroids’, ‘hydrocortisone’, ‘prednisolone’, ‘sulfa- salazine’, ‘mesalazine’, ‘olsalazine’, ‘balsalazide’, ‘az- athioprine’, and ‘mercaptopurine’. We also identified further trials from lists of references and expert reviews, including the Cochrane Library. We obtained and reviewed all references and deduced placebo and drug- induced remission rateswhere possible.We only included ‘intention-to-treat’ data, and used symptomatic patient based endpoints as a judge of induction of remission or maintenance, rather than artificial scoring systems or endoscopic gradings, as we felt this was the information that patients would wish to know. In some situations it was possible to include data on clinical improvement (without necessarily induction of remission) and we included this as we felt it would be useful. For each study we calculated an expected NNT by subtracting the placebo response rate from the drug response rate and dividing 100 by the percentage difference (95% confid- ence intervals were estimated using the Newcombe method). We have separated results by drug, induction or maintenance of remission and, in some cases, by dosage (e.g. low vs. standard or high dose 5-ASA). RESULTS Corticosteroids We could only identify two placebo-controlled trials of corticosteroids in the treatment of ulcerative colitis (Table 1).2, 3 Both studies were published over 40 years ago and one2 uses a form of corticosteroid (cortisone) not prescribed in current practice. To provide further information on the efficacy of corticosteroids we have examined data on the course or ‘natural history’ of ulcerative colitis in the precorticosteroid era, and other nonplacebo-controlled data. Before the introduction of corticosteroids in the treatment of ulcerative colitis in the early 1950s patients received supportive medical treatment inclu- ding high-calorie low-residue diet, vitamin supplements, superficial psychotherapy and, on occasion, psychiatric evaluation and treatment in the event of serious emotional disturbance.4 Mortality rates were high in the early 20th century ranging from 8 to 55%.5 In a review of 129 patients treated at the Radcliffe Infirmary, Oxford, between 1938 and 1948, 22% of patients with their first attack died.6 Interestingly, 69% of newly presented colitic patients in this study, who did not receive the treatment of the day (penicillin and/or sulphonamides), were symptom-free or had a ‘fair response’ with supportive measures alone. This is perhaps the best data we will ever have on the untreated course of ulcerative colitis at initial presen- tation. Further follow-up of patients treated in Oxford7 after 1953 (when corticosteroids were first being used) showed that patients in their first attack were much less likely to die (6.9% with treatment, 27.3% without) if they had received a specific medical treatment (such as adrenocorticotrophic hormone or corticosteroids). Current management of severe attacks of ulcerative colitis involves inpatient admission, i.v. hydrocortisone and, often, prednisolone enemas. There are no placebo- controlled trials of these treatments but nonplacebo- controlled trials from the 1970s8, 9 show remission Table 1. Studies of corticosteroids vs. placebo in the treatment of active ulcerative colitis Study Drug Treatment duration Placebo remission rate Drug remission rate Number needed to treat (95% CI) Truelove 19542 moderate-severe colitis Cortisone 100 mg initial dose orally 6 weeks 16/101 (15.8%) 45/109 (41.3%) 4 (3–7) Lennard-Jones 19603 mild colitis Prednisone 40–60 mg 3–4 weeks 3/18 (17%) 13/19 (68%) 2 (1.4–5) 144 J. R. BEBB & B. B. SCOTT � 2004 Blackwell Publishing Ltd, Aliment Pharmacol Ther 20, 143–149 rates of 60–73% with 75–81% either improved or in remission. Similar results have been obtained in other centres.10 Inpatient admission for i.v. hydrocortisone is not commonly required for exacerbations of UC, and many patients with mild to moderate exacerbations can be managed as outpatients with oral prednisolone. There are no placebo-controlled trials of oral prednisolone but, from placebo-controlled trials using prednisone (in equivalent dosages3) and noncontrolled studies,11 we can expect a remission rates of at least 65% with doses of at least 40 mg. Aminosalicylates (5-ASA preparations) Induction of remission in active colitis. We identified nine placebo-controlled trials of 5-ASA compounds in the treatment of active ulcerative colitis,12–20 (Table 2) helped by the Cochrane Library.21 None studied sulfa- salazine. Over 1200 patients have been included in these studies. A variety of preparations have been used in differing doses, but all in patients with mild to moderately active disease. For some of the studies data is available on improvement or remission, as opposed to induction of remission alone. For induction of remission we calculated an expected placebo remission rate of 10%, with 5-ASA compounds inducing remission in only 20%, giving an NNT of 10 (95% CI 7–21). When data including clinical improvement was added these rates improve to 34% for placebo and 58% for 5-ASA giving an NNT of 4 (95% CI 3–6). When data was split into what would now be considered low dose 13, 18, 19 and standard or high dose, little difference is made to the NNT. These calculations confirm our experience that 5-ASA alone is usually unsatisfactory in inducing remission in active disease. Maintenance of remission in ulcerative colitis. We identi- fied eight placebo-controlled trials of 5-ASA in main- tenance of remission in ulcerative colitis,22–29 including three using sulfasalazine (Table 3). Follow-up was between 6 months and 1 year. Thirty-eight per cent of Table 2. Studies of 5-ASA vs. placebo in the treatment of active ulcerative colitis Study Drug Duration (weeks) Placebo Drug Number needed to treat (95% CI) Remission (%) Remission or improvement (%) Remission (%) Remission or improvement (%) Remission Remission or improvement Hetzel 198612 Olsalazine 2 6 2/15 (13) 6/15 (40) 4 Schroeder 198713 Asacol 6 2/38 (5) 7/38 (18) 1.6 g 1/11 (9) 3/11 (27) 26 11 4.8 g 9/38 (24) 28/38 (74) 5 2 Robinson 198814 Olsalazine 3 g 4 14/48 (29) 21/50 (42) 8 Feurle 198915 Olsalazine 2 g 4 24/53 (45) 27/52 (52) 14 Zinberg 199016 Olsalazine 3 g 4 2/8 (25) 4/7 (57) 3 Sutherland 199017 Rowasa 6 8/44 (18) 2 g 8/45 (18) – 4 g 21/47 (45) 4 Sninsky 199118 Asacol 6 2/52 (4) 10/52 (19) 1.6 g 6/53 (11) 19/53 (36) 13 6 2.4 g 6/53 (11) 21/53 (40) 13 5 Hanauer 199319 Pentasa 8 11/90 (12) 49/90 (54) 1 g 19/92 (21) 65/92 (71) 12 6 2 g 28/97 (29) 77/97 (79) 6 4 4 g 28/95 (29) 80/95 (84) 6 3 Hanauer 199620 Olsalazine 12 12/90 (13) 2–2.9 g 11/92 (12) – ‡3 g 16/91 (18) 24 Overall All doses 27/270 (10) 116/348 (33) 124/622 (20) 380/655 (58) 10 (7–21) 4 (3–5) <2 g 15/180 (8) 66/180 (37) 26/156 (17) 87/156 (56) 12 (6–79) 5 (3–12) 2–2.9 g 25/232 (11) 93/254 (37) 45/242 (19) 139/262 (53) 13 (7–70) 6 (4–13) ‡3 g 25/218 (11) 80/228 (35) 53/224 (24) 154/237 (65) 8 (5–20) 3 (3–5) SYSTEMATIC REVIEW: DRUG EFFECTIVENESS IN ULCERATIVE COLITIS 145 � 2004 Blackwell Publishing Ltd, Aliment Pharmacol Ther 20, 143–149 patients on placebo can be expected to be in remission after this time period as opposed to 56% on 5-ASA, giving an NNT of 5 (95% CI 4–8). Higher remission rates were achieved in the original trials using sulfasal- azine, but this benefit (compared with more modern preparations) did not reach statistical significance. Azathioprine/6MP Azathioprine is increasingly used by gastroenterologists in patients requiring frequent courses of corticosteroids for ulcerative colitis. However, unlike in Crohn’s disease, where there is good evidence that azathioprine promotes both induction and maintenance of remis- sion,30, 31 there are few such data in UC. Jewell and Truelove’s study in 197432 randomized patients with active colitis to standard therapy plus azathioprine or dummy. They found no significant difference in remis- sion rates at 1 month between the two groups, or even at 1 year. However, when they split the patients into those having their first attack of colitis and those having a subsequent relapse, the relapse group were more likely to be in remission at one year if taking azathioprine (this almost reached statistical significance, P ¼ 0.055). Two other subsequent studies 33, 34 showed that azathio- prine had a steroid-sparing effect but that it did not alter the clinical or inflammatory course of the disease. Hawthorne et al. examined azathioprine in UC from a different angle.35 In a double-blind, placebo-controlled study they randomized outpatients with stable disease on azathioprine for at least 6 months to stop the drug or continue. The relapse rate in the group that stopped azathioprine was significantly higher than in the group that continued the drug. In further favour of a benefit of azathioprine was the median dose which was only 100 mg, implying that probably at least half the group were on <2 mg/kg body weight, and that the benefits of azathioprine may be even greater than they found. For these various reasons it is not possible to calculate meaningful NNTs. Ciclosporin Ciclosporin is the latest commonly used treatment in the management of (severe) ulcerative colitis. There is only one placebo-controlled trial,36 with small patient num- bers, which gave impressive response rates (Table 4). It would be unethical to expect further placebo-controlled trials in this situation given the known high mortality of untreated severe colitis, and the recognized benefits of corticosteroids. Therefore our ‘best guess’ for the expec- ted response rates to ciclosporin is based on nonplacebo- controlled data, which gives short-term response rates of 64–83%,37–39 similar to the original paper. Table 3. Studies of 5-ASA preparations vs. placebo in maintenance of remission of ulcerative colitis Study Drug Follow-up Placebo maintenance of remission rate (%) Drug maintenance of remission rate (%) Number needed to treat (95% CI) Misiewicz 196522 Sulfasalazine 2 g 1 year 8/38 (21) 24/42 (57) 3 Dissanayake 197323 Sulfasalazine 2 g 6 months 14/31 (45) 29/33 (88) 2 Riis 197324 Sulfasalazine variable dose 6 months 17/24 (71) 19/25 (76) 20 Sandberg 198625 Olsalazine 1 g 6 months 27/49 (55) 40/52 (77) 5 Wright 199326 Olsalazine 2 g 1 year 16/52 (31) 18/49 (37) 17 Miner 199527 Pentasa 4 g 1 year 34/102 (33) 59/103 (57) 4 Hanauer 199628 Asacol 6 months 25/87 (29) 0.8 g 40/90 (44) 6 1.6 g 38/87 (44) 7 Hawkey 199729 Mesalazine 1.6 g 6 months 45/111 (41) 59/99 (60) 5 Overall 186/494 (38) 326/580 (56) 6 (4–8) Sulfasalazine 39/93 (43) 72/100 (72) 3 (2–6) Other 5-ASA 147/401 (37) 254/480 (53) 6 (4–10) Table 4. Study of ciclosporin vs. placebo in the treatment of ulcerative colitis Study Drug Follow-up Placebo remission rate Drug remission rate Number needed to treat (95% CI) Lichtiger 199436 Ciclosporin 4 mg/kg per day i.v. 7 days 0/9 (0%) 9/11 (82%) 1.2 (1–2.5) 146 J. R. BEBB & B. B. SCOTT � 2004 Blackwell Publishing Ltd, Aliment Pharmacol Ther 20, 143–149 DISCUSSION Our aim when writing this review was to provide information on expected benefits of commonly used drugs in ulcerative colitis over placebo or no treatment. Our findings are summarized in Table 5, which could be useful as reference in the outpatient clinic. In the outpatient management of flares of ulcerative colitis corticosteroids are effective in 65% or more of patients. 5-ASA compounds are less effective, associated with a remission induction of only 20% and with an NNT of 10 (95% CI 7–21). They seem to be more effective if the daily dose is 3 g or greater (NNT ¼ 8; 95% CI 5–20). Sulfasalazine, although used for the longest time, has not been subjected to placebo- controlled trials of remission induction. However, it seems to be as effective as the other 5-ASA compounds with which it has been compared in controlled trials. 5-ASA compounds are of greater, although still limited, effectiveness in maintenance therapy with an NNT of 6 (95% CI 4–8). There is evidence suggesting a benefit of azathioprine in maintenance of UC and it has an additional steroid-sparing effect. In the inpatient management of severe attacks of ulcerative colitis, i.v. fluids and corticosteroids are associated with remission in up to three-quarters of patients. Untreated severe attacks are associated with high mortality. Ciclosporin is a useful drug in those that have failed to improve on i.v. corticosteroids and induces short-term remission in up to 80% of patients. As a result of this analysis, we would offer steroids to patients severely or moderately ill with active UC. If only mildly ill, we would first offer a 5-ASA preparation in high dosage (‡3 g daily) but change to steroids within a few weeks if there is no substantial improvement. When in remission, we would offer continued treatment with a 5-ASA preparation in the recommended doses, but would readily agree to stop it if a patient with distal disease felt it was not helpful – they would probably be correct. For patients with more extensive disease, who have an increased risk of colon cancer, we would encourage long term 5-ASA use because of the probable additional benefit of cancer prevention .40 If there is relapse within a year we would give a further course of steroids and start azathioprine (and continue long term if tolerated and effective). If the disease remains active we would try to achieve remission by inpatient treatment with i.v. steroids and/or ciclosporin. While we do not include it within this review, it is important to bear in mind that surgery remains an important management option in both the acute situation and chronic refractory ulcerative colitis, and in expert hands can yield excellent results. We have concentrated on placebo-controlled trials in this review so as to be able to advise patients of the benefits a treatment can offer them over no treatment. We accept that, for new treatments being developed, it may not be possible to conduct placebo-controlled trials, particularly in the acute setting with its high associated morbidity and mortality. We also accept that our review has limitations and could be criticized for having amalgamated trials with slightly different groups of patients, often on slightly different doses of similar drugs (especially in the case of the 5-ASA compounds). However, for the purpose of informing patients and doctors of likely benefit, such approximation is consid- ered acceptable. Table 5. Summary of expected placebo and drug induced remission/maintenance of remission rates in ulcerative colitis Drug Induction of remission or maintenance Placebo response rate % Drug response rate % Number needed to treat (95% CI) Prednisone in mild disease Induction of remission 17 68 2 (1.4–5) Prednisolone in mild to moderate disease Induction of remission – 65 – Hydrocortisone i.v. in severe disease Induction of remission – 60–73 – 5-ASA Induction of remission 10 20 10 (7–21) 5-ASA Induction of remission or improvement 35 59 4 (3–5) 5-ASA Maintenance 37 53 6 (4–8) Ciclosporin Induction of remission 0 82 1.2 (1–2.5) Ciclosporin Induction of remission (nonplacebo controlled trials) – 64–83 – SYSTEMATIC REVIEW: DRUG EFFECTIVENESS IN ULCERATIVE COLITIS 147 � 2004 Blackwell Publishing Ltd, Aliment Pharmacol Ther 20, 143–149 This review also discloses the severe limitations of our current tre