`
` Dig Dis 2009;27(suppl 1):68–75
` DOI: 10.1159/000268123
`
` The Placebo Response Rate in Irritable
`Bowel Syndrome and Inflammatory
`Bowel Disease
`
` Bruce E. Sands
`
` MGH Crohn’s and Colitis Center and Gastrointestinal Unit, Massachusetts General Hospital, Harvard Medical
`School, Boston, Mass. , USA
`
`72.37.250.188 - 11/6/2017 6:17:22 PM
`Reprints Desk
`
` Key Words
` Irritable bowel syndrome ⴢ Inflammatory bowel disease ⴢ
`Crohn’s disease ⴢ Ulcerative colitis ⴢ Placebo response
`
` Abstract
` The placebo response is the efficacy attributable to a treat-
`ment that is thought to have no specific pharmacologic ef-
`fect on the condition being treated. Although potentially
`helpful in clinical practice, high and unpredictable placebo
`response rates present a major impediment to the success of
`clinical trials in inflammatory bowel disease (IBD) and irrita-
`ble bowel syndrome (IBS). Diverse factors contribute to the
`placebo response rates observed in clinical trials. These in-
`clude patient characteristics, physician factors, frequency of
`study visits, characteristics of the outcome measures, con-
`comitant treatments, regression to the mean, properties of
`the intervention and treatment setting, timing of the prima-
`ry endpoint and natural history of the condition. Measures
`that may minimize the placebo response in IBD clinical trials
`include early timing of the primary endpoint, minimizing the
`number of study visits, restricting the patient population to
`those with documented inflammation (such as elevated bio-
`markers of inflammation or evidence of mucosal inflamma-
`tion), including patients with more severe symptoms (i.e.
`greater disease activity) and enrolling patients with prior
`failure of immune modulators or biologics. Attempts to lim-
`it the placebo response in IBS studies have proven more dif-
`
`ficult. Factors associated with higher placebo response rates
`in IBS studies include longer duration of treatment, greater
`number of office visits, frequency of administration of study
`intervention and overall treatment effect of the active agent
`under study. In the future, improved understanding of the
`factors that drive the placebo response rate should lead to
`more efficient study design and drug development.
` Copyright © 2010 S. Karger AG, Basel
`
` Introduction
`
` The ability of the human body to respond to inert
`stimuli in a positive fashion has long been recognized in
`Western medicine. A placebo is defined as a therapy be-
`ing given for a specific condition without any specific
`known activity for that condition. Although useful in the
`practice of medicine, in which physicians routinely call
`upon their skills in persuading the patient to a state of
`wellness, the placebo effect may be detrimental to the
`conduct of clinical trials in gastroenterology.
` Placebo treatment is usually required to maintain
`blinding in clinical trials. Meta-analyses of randomized
`control trials concerning many different conditions
`clearly demonstrate that inadequate concealment of al-
`location is a major factor contributing to biased results,
`leading to overestimation of treatment effect [1] . Al-
`though it is technically possible to conceal allocation by
`
`Fax +41 61 306 12 34
`E-Mail karger@karger.ch
`www.karger.com
`
` © 2010 S. Karger AG, Basel
`0257–2753/09/0275–0068$26.00/0
`
` Accessible online at:
`www.karger.com/ddi
`
` Bruce E. Sands, MD, MS
` Gastrointestinal Unit, Massachusetts General Hospital
`55 Fruit St., GRJ719
` Boston, MA 02114 (USA)
` Tel. +1 617 726 7411, Fax +1 617 726 3766, E-Mail bsands @ partners.org
`
`

`

`72.37.250.188 - 11/6/2017 6:17:22 PM
`Reprints Desk
`Downloaded by:
`
`randomizing to two active therapies, this approach does
`not permit measurement of the treatment benefit attrib-
`utable to the placebo response itself and also necessitates
`increasing the sample size requirements to create a study
`with sufficient power to discern treatment effect over an
`active agent. Arguments in favour of active comparators
`neglect to account for the possibility that there may be
`multiple active therapies which may be of similar efficacy,
`but which may differ primarily in their safety, cost or tol-
`erability. The application of non-inferiority studies de-
`signs also requires a relatively large sample size. There-
`fore, randomized controlled trials which incorporate a
`placebo treatment arm have become the standard meth-
`odology to establish the efficacy of a drug.
` Nevertheless, the introduction of a placebo arm creates
`a somewhat unpredictable variable in the conduct of ran-
`domized control trials. There has been some dispute about
`whether placebo treatments possesses any genuine clini-
`cal effects. A meta-analysis of studies in many different
`diseases and conditions has suggested that with regard to
`binary outcomes, the placebo response is minimal [2] .
`However, studies confined to single conditions clearly
`demonstrate the efficacy of placebo. Moreover, studies
`that incorporate a placebo arm tend to demonstrate pla-
`cebo response rates in excess of the outcomes noted in the
`natural history of the disease, suggesting effects of placebo
`above and beyond those observed without any treatment
`whatsoever [3] . Furthermore, it is clear that placebo treat-
`ments themselves vary with regard to the response rates
`and are not identical in the strength of their effects [4] .
` It is useful, as well, to distinguish between the ‘placebo
`effect’ and the ‘placebo response’ [5] . The placebo re-
`sponse is the individual reaction that arises to placebo
`and which may result from a variety of features. Patient
`and physician expectations may vary and may serve to
`heighten the effect of the placebo response. In addition,
`operant conditioning may permit transfer of benefit from
`an active drug to placebo, a feature which may invalidate
`crossover study design for some conditions. Further-
`more, the power of the physician’s suggestions about re-
`sponse to the intervention may contribute to placebo re-
`sponse. It is clear that in some situations there is a neuro-
`biological basis to the placebo response. For example, in
`studies of chronic pain, it has been noted that patients
`who respond to a placebo therapy may have abrogation of
`their analgesia with treatment of the opiate antagonist
`naloxone [6] . However, the neurobiological basis of the
`placebo response is incompletely understood and may
`not always be a factor in an individual’s response to pla-
`cebo or fully explanatory for the entirety of the placebo
`
`response. Variability in the strength of these various con-
`tributions to the placebo response will also lead to vari-
`ability in the placebo response noted in an individual.
` In aggregate, the effect of the true placebo response
`contributes to the placebo effect that is observed in a group
`of subjects and which may affect the results of the clinical
`trial [4] . In addition to the true placebo response, an amal-
`gamation of the placebo responses seen in all the individ-
`uals assigned to placebo, other factors may contribute to
`the clinical response in patients assigned to the placebo
`arm. These include regression to the mean and the natural
`history of the disease that is being treated [3] . Regression
`to the mean refers to the tendency for patients with outly-
`ing values of disease activity to return over time to a more
`moderate value that approximates the average value of the
`group under study. This is especially important in studies
`where the inclusion criteria select for patients with high
`levels of disease activity, a situation where it is more likely
`that a patient will gradually improve rather than deterio-
`rate further. With regard to natural history, diseases that
`are waxing and waning in nature, such as inflammatory
`bowel disease (IBD), may be subject to improvement over
`the course of a study. Contributing to the observed natural
`history in the context of the study may be the effects of
`permitted concomitant medications. Particularly if ade-
`quate wash-in or wash-out periods are not observed prior
`to enrollment, such drug effects may contribute to the ap-
`parent benefits observed in the placebo arm:
` Kaptchuk [7] described five components of the place-
`bo response in the context of clinical trials:
` (1) Patient characteristics, including features such as
`expectation, adherence to therapy and perception of con-
`trol.
` (2) Characteristics of the practitioner, who may deliv-
`er a positive or negative suggestion about response and
`who may be more or less effective in communication of
`the potential benefits of the intervention.
` (3) The interaction between the patient and the prac-
`titioner. This includes features such as the frequency of
`contacts during the study, the duration of the study and
`the development of trust between the patient and the
`practitioner over the course of time.
` (4) The nature of the illness itself. This includes the
`course of the disease and the physiology of the illness
`along a functional and organic spectrum. This is par-
`ticularly important given the potential contributions of
`mood disorders, such as depression, to a greater percep-
`tion of illness in patients with such co-morbid condi-
`tions. To the extent that pain and general well-being may
`be affected by concomitant mood disorder, patients may
`
` Placebo Response Rate in IBS and IBD
`
`Dig Dis 2009;27(suppl 1):68–75
`
`69
`
`

`

`72.37.250.188 - 11/6/2017 6:17:22 PM
`Reprints Desk
`Downloaded by:
`
`be more or less subject to higher or lower placebo re-
`sponse rates.
` (5) The treatment and the setting in which the treat-
`ment is given. This includes characteristics such as the
`efficacy of the active drug being compared to the placebo
`and the properties of the intervention or agent itself.
`Some studies have noted an increased placebo response
`rate in studies where the active drug has known efficacy
`and is expected to be fairly effective. In addition, factors
`such as the appearance and packaging of the placebo, in-
`cluding the colour, brand name, frequency of use and
`mode of administration all may contribute to the placebo
`response rate .
`
` Irritable Bowel Syndrome
`
` Studies of the placebo response rate in irritable bowel
`syndrome (IBS) underscore the difficulties of predicting
`and limiting this effect in clinical trials. In 2005, Pitz et
`al. [8] performed a systematic review of placebo-con-
`trolled trials in patients with IBS to assess the variables
`that correlated with increasing placebo response. They
`explored studies that assess global symptoms improve-
`ment or improvement in abdominal pain as outcomes.
`The authors found 84 trials meeting their criteria for in-
`clusion, including 53 that assessed global symptoms im-
`provement, 39 that assessed decreased abdominal pain
`and 8 that assessed both definitions of response. Overall,
`global response rates were noted to occur at a mean of
`52% and a range of 46–53%. The mean response rate for
`abdominal pain was 38%, with a range of 29–39%. The
`authors examined numerous variables as potential pre-
`dictors of the placebo response. These included frequen-
`cy of intervention, overall treatment effect, duration of
`study, patient age, duration of run-in period, year of pub-
`lication, male-to-female ratio, the number of patients in
`the treatment group or placebo group, the number of
`doctor visits, the number of other visits and the number
`of phone contacts. With regard to global improvement,
`the frequency of intervention was found to be significant-
`ly associated with placebo response on multivariate anal-
`ysis. In addition, the overall treatment effect of the active
`agent was also moderately associated with global im-
`provement in the placebo arm. The duration of the study
`was also associated with global improvement on univari-
`ate analysis, but did not prove to be an independent pre-
`dictor of global improvement. Finally, parallel study de-
`sign was a significant predictor of global improvement
`response in multivariate analysis.
`
` Pitz et al. [8] also identified predictors of placebo re-
`sponse rate using abdominal pain as the outcome. Fre-
`quency of intervention and the overall treatment effect of
`the active agent were positively associated with the pla-
`cebo response rate and were significant on multivariate
`analysis. Year of study, median age and duration of study
`run-in period were negatively correlated with the placebo
`response rate for abdominal pain, but were not indepen-
`dent predictors on multivariate analysis.
` A second meta-analysis published in 2005 by Patel et
`al. [9] looked at published, English language randomized
`control trials with 20 or more IBS patients who were
`treated for at least 2 weeks. The authors limited their
`analysis to studies looking at global response or overall
`improvement in symptoms. The authors considered po-
`tential factors such as study design, study duration, use of
`the run-in phase, Jadad score, entry criteria, number of
`office visits, number of office visits per study duration,
`use of diagnostic testing, gender, age and type of medica-
`tion studied as potential predictors for placebo response.
`In all, the authors found 45 placebo-controlled random-
`ized controlled trials that met their inclusion criteria. The
`global response rates ranged from 16 to 71% with a popu-
`lation weighted average of 40.2% (95% CI: 35.9–44.4%).
`Lower placebo response rates were associated with fulfill-
`ment of Rome criteria for a study entry and an increased
`number of office visits. The authors suggested that main-
`taining more stringent entry criteria and increasing the
`number of office visits would serve to decrease the pla-
`cebo response rate in future studies.
` Dorn et al. [10] performed a meta-analysis of placebo
`response in complementary and alternative medicine
`(CAM) trials in IBS. The authors performed a systematic
`review and meta-analysis of randomized placebo-con-
`trolled trials of CAM therapies for IBS performed be-
`tween 1970 and 2006. Global symptom improvement was
`assessed in all studies. Nineteen studies met the authors’
`inclusion criteria. The pooled estimate of the placebo re-
`sponse rate was 42.6% (95% CI: 38.0–46.5%). Higher pla-
`cebo response rates correlated with the longer duration
`of treatment and greater number of office visits, in con-
`trast to the findings of Patel et al. [9] . Although the pla-
`cebo response rates in CAM trials of IBS were high, it was
`noted that the magnitude of the placebo response rate in
`these studies was quite similar to that seen in studies of
`conventional medications. It is worth noting that each of
`these meta-analyses adopted the practice common in IBS
`studies of defining the treatment outcome as global re-
`sponse, i.e. improvement in symptoms rather than reso-
`lution of symptoms as has been suggested in some studies
`
`70
`
`Dig Dis 2009;27(suppl 1):68–75
`
` Sands
`
`
`
`

`

`72.37.250.188 - 11/6/2017 6:17:22 PM
`Reprints Desk
`
`Recruitment through gastroenterology practice
`
`Study Enrollment: consent process, background and baseline
`data collection, schedule appointments for acupuncture
`
`Randomization to Phase I
`
`Augmented Placebo Context
`& Sham Acupuncture
`
`Limited Context
`& Sham Acupuncture
`
`Minimal Context Wait List
`
`Main outcome data collection & re-randomization to Phase II
`
`Augmented
`Context &
`Verum
`Acupuncture
`
`Limited
`Context &
`Verum
`Acupuncture
`
`Augmented
`Context &
`Sham
`Acupuncture
`
`Limited
`Context &
`Sham
`Acupuncture
`
`Minimal
`Context &
`Wait List
`
`Post-treatment data collection before unblinding
`
`Fig. 1. Outline of the study design to define
`components of the placebo effect in a study
`of acupuncture in IBS, adapted from [12].
`
`in IBS, which may contribute to a decrease in placebo re-
`sponse rate.
` An elegant study has been performed that attempts to
`dissect various components of the placebo effect in ran-
`domized controlled trials in IBS. Kaptchuk et al. [11] pub-
`lished in 2008 a randomized controlled trial where they
`sought to investigate whether placebo effects can be sepa-
`rated into three potential components: (1) effects attribut-
`able to patient assessment and observation, (2) effects due
`to the therapeutic ritual of the placebo treatment, and (3)
`effects arising from a supportive patient-practitioner rela-
`tionship. The authors performed a 6-week, single blind
`three-arm randomized controlled trial. The interventions
`were assignment to a waiting list (observation), placebo
`acupuncture alone (limited treatment) and placebo acu-
`puncture with patient-practitioner relationship augment-
`ed by warmth, attention and confidence (so-called ‘aug-
`mented’ intervention). The outcomes included measure-
`ment of global improvement on a Likert scale ranging
`from 1 to 7, measurement of adequate relief of symptoms,
`symptom severity score and quality of life. After recruit-
`ment through gastroenterology practices and informed
`consent, patients were scheduled for appointments for
`acupuncture and were randomized to phase I of the study.
`Patients were randomized to augmented placebo context
`with sham acupuncture, to limited placebo context and
`sham acupuncture, or to the minimal context wait list.
`After 3 weeks, the data were collected and there was an
`additional re-randomization; patients assigned to receive
`
`acupuncture were re-assigned to receive either augment-
`ed context and true acupuncture, limited context and true
`acupuncture, augmented context and sham acupuncture,
`or limited context and sham acupuncture, while patients
`assigned to the wait list continued in this arm ( fig. 1 ). In
`all, 262 adult patients were enrolled, including 76% wom-
`en. The mean age was 39 years with a standard deviation
`of 14 years. All patients had IBS diagnosed by the Rome II
`criteria and a score of 6 150 on the Symptom Severity
`Scale. At 3 weeks, the results demonstrated a semi-addi-
`tive effect of increasingly intensive placebo treatment. Pa-
`tients who were assigned to the waiting list experienced
`some improvement in each of the measured outcomes.
`However, patients assigned to limited context of sham
`acupuncture alone noted significantly greater improve-
`ment in each of the outcomes when compared to the wait
`list arm, while patients assigned to the augmented arm
`noted still greater increase. The p value for the trend for
`each outcome was ! 0.001 along the entire spectrum of
`treatments and for each measured outcome. The authors
`concluded that various factors contribute to the placebo
`effect and may be combined with one another in a manner
`approximating dose escalation of the components. Fur-
`thermore, the effects observed with placebo therapy were
`not only statistically significant but also clinically mean-
`ingful. These results suggest the potential utility of the
`placebo response in the treatment of individual patients,
`but also suggest the difficulties in controlling for placebo
`effect in randomized controlled trials.
`
` Placebo Response Rate in IBS and IBD
`
`Dig Dis 2009;27(suppl 1):68–75
`
`71
`
`

`

`of eight variables which include four measures of symp-
`toms, one factor for the use of anti-diarrhoeal agents, two
`points of physical examination (abdominal mass and
`body weight) and one laboratory value (hematocrit) ( ta-
`ble 1 ). The symptoms that contribute to the index include
`the number of liquid or soft stools as recorded each day
`for a period of 7 days, abdominal pain ratings over the
`course of a week, and ratings of general well-being. In ad-
`dition, complications such as arthritis or arthralgia, iritis
`or uveitis, erythema nodosum, and pyoderma gangreno-
`sum also contribute points to the score [14] . It has been
`observed that the major contribution to the variability of
`the CDAI arises from the first three symptoms, diar-
`rhoea, abdominal pain and general well-being, whereas
`the other components, while adding points to a score and
`increasing the severity, are less responsive overall to
`changes over relatively short periods of time such as one
`or two months, which is typical for an induction trial [15] .
`In addition, it is worth noting that the variables of ab-
`dominal pain and general well-being contribute signifi-
`cantly to the overall score, although these particular
`symptoms are highly subjective and may be affected to a
`greater extent by the placebo response. Consequently,
`nothing about the CDAI is entirely specific for Crohn’s
`disease itself and if administered to a patient with symp-
`tomatic IBS, the CDAI score would register in the ‘active
`disease’ range.
` Noting the difficulties presented by high placebo re-
`sponse rates in clinical trials of Crohn’s disease, Su et al.
` [16] conducted a meta-analysis to explore factors that
`may be associated with higher placebo response rates.
`The authors considered 30 studies that fulfilled their in-
`clusion criteria. Twenty-one of these included the use of
`the CDAI for definitions of response or remission. Ap-
`plying a definition of remission as a CDAI score of either
` ! 150 or ^ 150, the pooled estimate for clinical remission
`was 18% (95% CI: 14–24%, overall range: 0–50%). There
`was significant heterogeneity among the different stud-
`ies. Applying a response definition of a decrease by 6 100
`points from baseline, the pooled estimate for response
`rates was 19% (95% CI: 13–28%, range: 0–46%). Again,
`significant heterogeneity was observed for this outcome.
`The authors also noted that placebo remission rates
`tended to increase over the period of observation as study
`duration increased. Considering a large variety of pos-
`sible characteristics that may have contributed to place-
`bo remission rates, the authors found that the adminis-
`tration of an injectable medication led to a lower remis-
`sion rate, whereas longer duration of follow-up was
`associated with a higher placebo remission rate. Interest-
`
`72.37.250.188 - 11/6/2017 6:17:22 PM
`Reprints Desk
`Downloaded by:
`
` Sands
`
`
`
`Table 1. Variables of the CDAI [14]
`
`Variable description
`
`Vari-
`able
`
`Multi-
`plier
`
`1
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`8
`
`!7
`
`!20
`
`number of liquid or soft stools (each day for 7 days) !2
`abdominal pain (0 = none, 1 = mild, 2 = moderate,
`!5
`3 = severe)
`general well-being (0 = generally well, 1 = slightly
`under par, 2 = poor, 3 = very poor, 4 = terrible)
`number of listed complications [arthritis or arthral-
`gia, iritis or uveitis, erythema nodosum or pyoderma
`gangrenosum or aphthous stomatitis, anal fissure or
`fistula or abscess, other fistula, fever over 37.8°C
`(100°F)]
`use of diphenoxylate or loperamide for diarrhoea
`(0 = no, 1 = yes)
`abdominal mass (0 = no, 2 = questionable,
`5 = definite)
`hematocrit [males: 47-Hct (%), females: 42-Hct (%)] !6
`body weight [(standard weight – actual body
`!1
`weight)/standard weight] !100 (add or subtract
`according to sign)
`
`!30
`
`!10
`
` Crohn’s Disease
`
` Over the last 15 years, numerous studies have been
`performed for therapeutic agents in Crohn’s disease. It is
`sobering to note that the majority of these studies have
`failed to demonstrate the benefit of agents previously
`thought to have had therapeutic benefit or a physiologic
`effect on the disease processes that underlie this condi-
`tion. The placebo response rates in clinical trials were
`observed to be highly variable and in many cases quite
`high. Although clearly an organic condition, it is equally
`clear that a functional component may contribute to the
`symptoms of Crohn’s disease. However, the practice of
`incorporating many different symptoms into a single dis-
`ease index number is a longstanding tradition in clinical
`trials in Crohn’s disease and ulcerative colitis. Conceptu-
`ally, the need for a composite disease activity index arises
`from the knowledge that the symptoms of Crohn’s dis-
`ease illness are quite variable from patient to patient, pre-
`senting a difficulty in comparing many different patients
`in a clinical trial. Since the late 1970s, the use of the
`Crohn’s Disease Activity Index (CDAI) has been the
`mainstream methodology for quantifying response to
`therapy in clinical trials in Crohn’s disease [13] .
` The CDAI was created for and validated by the Na-
`tional Cooperative Crohn’s Disease Study [14] . It consists
`
`72
`
`Dig Dis 2009;27(suppl 1):68–75
`
`

`

`72.37.250.188 - 11/6/2017 6:17:22 PM
`Reprints Desk
`Downloaded by:
`
`ingly, greater frequency of follow-up was associated with
`a lower placebo remission rate, while the total number of
`follow-up visits was associated with a slightly higher pla-
`cebo remission rate. There was a slight trend to a decreas-
`ing placebo response rate as the publication year in-
`creased, belying the conventional wisdom that placebo
`response rates have been gradually increasing in recent
`years. The authors also identified prior exposure to var-
`ious treatments as predictors of placebo response rates.
`In particular, concurrent corticosteroids or concurrent
`immune modulators were strongly associated with a
`lower placebo response rate (odds ratio: 0.47 for concur-
`rent corticosteroids and 0.48 for concurrent immune
`modulators). These findings suggest that patients with
`more refractory disease may experience lower placebo
`response rates. Interestingly, patients with prior cortico-
`steroid therapy had only a slight decrease in placebo re-
`sponse rate (odds ratio: 0.94). The authors also consid-
`ered characteristics of the enrolled patients. Patients with
`prior surgery had a slightly decreased placebo response
`rate (odds ratio: 0.78), while the effect of the baseline
`CDAI score was more notable. A mean or median CDAI
`score of the placebo group at entry of 6 250 versus ! 250
`was associated with an odds ratio of 0.5, implying that the
`placebo response is halved in studies that enroll patients
`with more active disease. Similarly, studies with a mini-
`mum inclusion CDAI score of 6 200 versus 6 150 had an
`odds ratio for a placebo response rate of 0.52. These find-
`ings again suggest that patients who are more ill upon
`entry and have more refractory disease experience a low-
`er placebo response rate. Also notable is the fact that no
`single factor accounted for all the heterogeneity in the
`placebo response rates of the studies, suggesting the com-
`plexity of engineering studies to minimize the placebo
`response rate. It was also found that factors that influ-
`enced the placebo response rates were similar to those
`affecting placebo remission rates in Crohn’s disease. The
`absolute benefit of active treatment beyond placebo treat-
`ment is generally larger when the outcome was response
`rather than remission.
` A large concern with regard to the placebo response in
`Crohn’s disease studies has been the inclusion of patients
`with minimal demonstrable inflammatory disease. It is
`possible that functional symptoms predominate in these
`patients and that such patients may be more subject to the
`same factors that contribute to the high placebo response
`rate seen in studies of IBS, as opposed to improvement of
`symptoms generated by inflammation in the gut.
` Such concerns date back to earlier studies of anti-
`tumour necrosis factor agents. In particular, studies of
`
`CDP870 (now called certolizumab pegol) heightened
`awareness of this possibility. In a phase II study, patients
`with active Crohn’s disease were assigned to placebo in-
`jections on a monthly basis or to one of three different
`doses of certolizumab pegol [17] . The primary endpoint
`was clinical response at week 12, defined as a decrease in
`the CDAI of 6 100 points from baseline or a CDAI of
` ^ 150 points. At week 2, the placebo response rate was
`15%, while the treatment arms had response rates rang-
`ing from 30 to 33%. These results were statistically sig-
`nificant. However, by week 12, the placebo response rate
`had risen to 36% compared to drug response rates rang-
`ing from 37 to 44%, none of which were significant in
`comparison to placebo. A post hoc analysis confined to
`those patients who had a baseline C-reactive protein of
` 6 10 mg/l appeared to demonstrate a notable decrease in
`the placebo response rate at week 12 to 18% as opposed to
`the previous value of 36%. This suggested that the pa-
`tients with active inflammatory disease had a beneficial
`effect from the drug, whereas patients without evidence
`of inflammation experienced primarily a placebo re-
`sponse. Subsequent post hoc analyses in other failed stud-
`ies in Crohn’s disease with other agents further enhanced
`these post hoc observations. In particular, analyses con-
`fined to patient populations with elevated C-reactive pro-
`teins and/or who had been previously treated with anti-
`tumour necrosis factor agents or immune modulators,
`appeared to have lower placebo response rates and when
`such post hoc analyses were applied, an apparent drug
`effect could be demonstrated.
` However, the strategy of confining analyses to patients
`with elevated C-reactive protein was undercut by pro-
`spective and predefined analysis in the PRECiSE 2 of cer-
`tolizumab pegol in active Crohn’s disease [18] . In this
`study, the primary analysis was confined to patients with
`elevated C-reactive protein 6 10 mg/l at baseline. At week
`26, the response and remission rates were virtually identi-
`cal for the high C-reactive protein population and for the
`full intent to treat population (all treated patients). De-
`spite this failure to demonstrate the benefits of prescreen-
`ing by C-reactive protein prospectively, it has become
`common practice to enroll patients in Crohn’s disease tri-
`als who have demonstrable inflammatory activity. This
`might include the demonstration of significant inflam-
`mation on colonoscopy, the presence of an elevated C-re-
`active protein or other inflammatory marker, or in some
`cases the presence of elevated stool markers of inflamma-
`tion, including faecal lactoferrin or calprotectin.
` In summary, studies in Crohn’s disease may best min-
`imize the placebo response rate when the primary end-
`
` Placebo Response Rate in IBS and IBD
`
`Dig Dis 2009;27(suppl 1):68–75
`
`73
`
`

`

`72.37.250.188 - 11/6/2017 6:17:22 PM
`Reprints Desk
`Downloaded by:
`
`point can be measured at an earlier rather than a later
`time point, preferably within 6 weeks of baseline, when
`patients with more refractory or more symptomatic dis-
`ease can be enrolled, and when patients with bona fide
`inflammation can be enrolled.
`
` Ulcerative Colitis
`
` In contrast to Crohn’s disease and IBS, the study of
`ulcerative colitis presents the opportunity for direct ob-
`servation of the disease process through the use of sig-
`moidoscopy or colonoscopy. A large number of disease
`activity indices exist for use in clinical trials and are de-
`scribed in the literature [19] . In general, indices that do
`not incorporate sigmoidoscopy appear to correlate well
`with those that do. In recent years, the de facto choice for
`clinical activity index in ulcerative colitis trials has been
`the Mayo Score [20] . This scoring system includes stool
`frequency, a rating of rectal bleeding, a patient’s rating of
`functional assessment, sigmoidoscopic appearance and
`the physician’s global assessment, the latter of which takes
`into account all of the above factors. All except for the
`subject’s functional assessment are included in a sum-
`mary score ranging from 0 to 12, with each individual
`score ranging from 0 to 3 [20] .
` In recent studies, the trend has been toward incorpo-
`rating a primary endpoint of remission, which includes
`objective evidence of normalization in mucosa on sig-
`moidoscopy. A meta-analysis by Su et al. [21] published
`in 2007 validates this approach. The authors performed
`a systematic review of the literature and identified 40
`studies that met inclusion criteria for meta-analysis. Ap-
`plying criteria for remission defined as an Ulcerative
`Colitis Disease Activity Index (UCDAI) score of 0, the
`authors calculated a pooled estimate of placebo remission
`rate of 13% (95% CI: 9–18%, range: 0–40%, median re-
`mission rate: 12%). A pooled estimate of the placebo re-
`sponse rate, defined variously as a decrease from baseline
`UCDAI of at least either two or three points, identified a
`placebo response rate of 28% (95% CI: 23–30%, range:
`0–67%, median: 30%). As with other disease states, more
`stringent definitions of remission or response yield lower
`placebo rates. Studies that defined remission as a UCDAI
`of 0 yielded a remission rate of 5% (95% CI: 2–16%),
`whereas studies defining remission as a UCDAI of ! 3 had
`a placebo remission rate of 17% (95% CI: 10–28%). The
`authors also identified higher placebo remission rates as-
`sociated with longer study duration,