The Biosimilar Nocebo Effect? A Systematic Review of Double-Blinded Versus Open-Label Studies

BACKGROUND: Several authors have hypothesized that adverse drug events (ADEs) upon switching from reference biologics to biosimilar products are related to the nocebo effect. However, a thorough and current review of the existing literature has not been conducted. OBJECTIVE: To evaluate if patient and/or physician knowledge of a switch from a reference biologic product to a biosimilar product was associated with an increase in ADEs likely to be susceptible to the nocebo effect. METHODS: Studies reporting efficacy and safety outcomes of a switch from a reference product to a biosimilar product were reviewed. Biologics with FDA-approved biosimilars in the United States were considered for review, including adalimumab, bevacizumab, etanercept, and infliximab. Studies were identified by searching controlled vocabulary (e.g., MeSH terms) and keywords within MEDLINE (via PubMed) and Embase. Descriptive statistics were used to quantify subjective and objective complications in double-blinded and single-blinded or open-label studies. RESULTS: Thirty-one trials including 3,271 patients were reviewed in the full analysis. Median discontinuation rates for any reason were 14.3% (range = 0.0-33.3) in open-label studies compared with 6.95% (range = 5.2-11.0) in double-blinded studies. Discontinuation rates for ADEs were 5.6% (range = 0.0-24.2) in open-label studies versus 3.1% (range = 2.0-5.2) in double-blinded studies, suggesting the nocebo effect does affect biosimilar adoption. Subgroup analyses of antidrug antibody (ADA) development and infusion reactions were similar between infliximab open-label and double-blinded studies. Discontinuation rates for any reason, for ADEs, and for lack of efficacy were generally higher in infliximab open-label trials compared with double-blinded trials. Etanercept biosimilar discontinuation rates for any reason were similar between study designs; however, incidences of injection site reactions and discontinuation rates for ADEs were higher in double-blinded compared with open-label study designs. CONCLUSIONS: Current evidence is insufficient to confirm a biosimilar nocebo effect, although higher discontinuation rates in infliximab biosimilar open-label studies support this theory. Further studies are needed to evaluate the existence of a biosimilar nocebo effect.

I n 2010, the Biologics Price Competition and Innovation (BPCI) Act created an abbreviated approval pathway, 351(k), for biosimilars. Biosimilars are biological products approved by the U.S. Food and Drug Administration (FDA) that have demonstrated a high degree of similarity and a lack of clinically relevant differences compared with a reference (brandname) FDA-approved biologic product. 1 Biosimilar products may present opportunities for substantial cost savings for health systems if used in the place of higher-cost reference biologic products. 2,3 By 2020, approximately $100 billion in cost per year of biologic products will lose patent exclusivity, which exemplifies the large cost savings potential. 4 Indeed, the National Health Service of the United Kingdom reported a cumulative cost savings of greater than 38 million pounds over a 2-year period, solely from the introduction of infliximab and etanercept biosimilars. 5 However, the use of biosimilar products in the United States has remained relatively slow, compared with Europe, since the passage of the BPCI Act. 6 Many challenges to the widespread adoption of biosimilars have been reported in the United States, including variable cost, insurance coverage, and reimbursement; aggressive reference product exclusivity contracting; lack of multiple biosimilar products for one reference product; and provider concerns over the strength of evidence supporting safety and efficacy of biosimilars. Clinical concerns seem especially prevalent when considering switching maintenance therapy to a biosimilar product in a currently stable patient. [6][7][8] Some authors have also suggested that the nocebo effect may further hinder the use of biosimilar products. 9,10 • The nocebo effect has been defined as negative expectations that lead to negative consequences following an inert exposure. • It has been hypothesized that adverse drug events (ADEs) of biosimilar switches are related to the nocebo effect without significant comparisons between open-label versus blinded outcomes.

What is already known about this subject
• By comparing subjective and objective outcomes in open-label and blinded trial designs, this comprehensive review evaluated the possibility of a nocebo effect when switching from originator to biosimilar products. • Current evidence is insufficient to confirm a biosimilar nocebo effect, although higher discontinuation rates in infliximab biosimilar open-label studies support this theory. • The nocebo effect may further inhibit biosimilar adoption, especially in light of previously established provider discomfort with switching to a biosimilar product.
greater than 95% of rheumatologist respondents stated that they would prefer prescribing a biologic reference product over its biosimilar product. 7 Up to 41% of patient respondents also reported reluctance to accept biosimilars due to potential side effects or long-term problems. 8 If this so-called nocebo effect is indeed present, it may further inhibit biosimilar uptake, especially in light of established provider discomfort and negative expectations. The Anglo-Scandinavian Cardiac Outcomes Trial was one of the first and largest studies to document an increase in ADEs from the nocebo effect. In the first phase of the trial, patients were randomized in a double-blinded fashion to atorvastatin or placebo. During the open-label extension phase, all patients were offered the option to receive atorvastatin therapy. The incidence of definite or probable statin-associated muscle The nocebo effect does not have a single consensus definition, but it may broadly be described as negative expectations that lead to negative consequences. In other words, it is the negative equivalent to the placebo effect. 9,11 Social observations, perceived dose, verbal suggestions of symptoms, and baseline symptom expectations are the strongest factors that may increase the risks of experiencing nocebo effects. 12 An increased incidence of adverse drug events (ADEs) related to the nocebo effect has previously been reported for several medications, including 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors (statins), finasteride, beta blockers, caffeine, and antiepileptics. [13][14][15][16][17] Surveys demonstrate physician hesitancy surrounding switching practices, potentially as a result of general knowledge gaps about biosimilar products. For example, in a survey,  (25.6)   During the open-label extension phase, however, significantly more SAMS were reported in the group receiving atorvastatin compared with the atorvastatin nonusers (HR = 1.41; 95% CI = 1.1-1.79). This result indicated that the patients' definitive knowledge of receiving statin therapy, and the associated potential side effects, may have increased the incidence of ADEs. 13 Indeed, the knowledge of the potential for negative effects may increase a patient's perception of ADEs. 18 One study reported that patients who were educated about potential medication-related side effects experienced symptoms almost 4 times as often as patients who were not educated on these symptoms. 15 Despite the evidence supporting a nocebo effect with traditional small-molecule medications, few studies have closely evaluated its presence in the context of biosimilar products. A recent open-label study of patients who were switched from the reference product to an infliximab biosimilar, CT-P13, revealed that the majority of treatment discontinuations were related to subjective ADEs, such as arthralgia, fatigue, pruritus, and myalgia. The authors suggested that these ADEs and subsequent treatment discontinuations were attributed to the  2017)   The Biosimilar Nocebo Effect? A Systematic Review of Double-Blinded Versus Open-Label Studies nocebo effect and not related to the efficacy or safety of the biosimilar product. 10 Similarly, a recent small observational study evaluated a nocebo-effect response defined as unexplained, undesirable therapeutic effects after a switch from a reference product to an infliximab biosimilar. 19 An overall nocebo response rate of 12.8% (16/125) resulted, further suggesting a negative effect on patients' perceived disease burden.
A thorough and current review of the existing literature has not been conducted. We hypothesized that the incidence of subjective ADEs (e.g., patient discomfort) would be higher in openlabel biosimilar switching studies, as they would be more likely to be affected by a nocebo effect. Conversely, the incidence of objective ADEs (e.g., laboratory values) would remain consistent between open-label and double-blinded studies, as they would be less likely to be affected by a nocebo effect. We categorized ADEs as highly likely, intermediate, and unlikely to be affected by the nocebo effect. Discontinuation rates were considered as highly likely, infusion and injection site reactions as intermediate, and antidrug antibody (ADA) development as unlikely. Therefore, the objective of this review was to evaluate if patient and/or physician knowledge of a switch from a reference biologic product to a biosimilar product was associated with an increase in ADEs likely to be susceptible to the nocebo effect.

■■ Methods
Studies reporting efficacy and safety outcomes of a switch from a reference product to a biosimilar product were reviewed. Primary literature that focused on FDA-approved biosimilars was considered for review, including adalimumab, bevacizumab, etanercept, and infliximab. Studies of filgrastim were excluded due to the generally low immunogenicity potential and low incidence of discontinuations related to ADEs. The research protocol and initial search was completed before FDA approval of trastuzumab-dkst, so it was not included in this review. Studies were identified by searching controlled vocabulary (e.g., MeSH terms) and keywords within MEDLINE (via PubMed) and Embase. The last search was conducted on A standardized data collection sheet was used to extract data from each trial, including study design, patient population, baseline characteristics, outcome measures, and duration of follow-up. Descriptive statistics were used to quantify results. Subjective and objective complications were compared between double-blinded and single-blinded or open-label studies.

■■ Results
For inclusion in the review, 1,153 results were assessed: 591 for infliximab, 228 for adalimumab, 198 for etanercept, and 117 for bevacizumab. After removing duplicates and assessing relevance of outcome measures, 31 trials including 3,271 patients were reviewed in the full analysis. Only 1 relevant trial was identified that included a switch to adalimumab biosimilar products, so it was excluded from analysis. No relevant studies were identified for bevacizumab.
Of the 31 included trials, 28 involved switches from infliximab and 3 from etanercept to their biosimilar product counterparts. Trials were primarily conducted in the United States and United Kingdom. The most common indications for treatment were inflammatory bowel disease (IBD), rheumatoid arthritis (RA), and psoriasis (15, 6, and 2 studies, respectively). Six studies included multiple disease states. The median duration of follow-up after the switch to the biosimilar product was 48 weeks (range = 16-80 weeks). A summary of trial design and baseline characteristics for infliximab and etanercept studies is  Griffiths et al. (2017) 48 [48][49][50] reported in Table 1 and Table 2. A summary of individual study results is reported in Table 3 and Table 4, respectively.

Baseline Characteristics of Etanercept Biosimilar Switching Studies
Twenty-eight studies involving a total of 2,956 patients switched from infliximab biologic products to biosimilars were included in the review.  Two of these studies were doubleblinded randomized controlled trials (RCTs). Presence of baseline ADAs was reported in 12 infliximab open-label studies. ADA development and infusion reactions were similar between open-label and double-blinded studies. Discontinuation rates for any reason, for ADEs, and for lack of efficacy were generally higher in open-label trials compared with double-blinded trials. A summary of reported discontinuation rates and ADEs for infliximab studies is reported in Table 5.
Three studies involved switching 315 patients from etanercept to a biosimilar product. [48][49][50] Two publications evaluated different time points of the same double-blinded RCT. One study was a prospective open-label design. 50 Of note, the 2 double-blinded RCTs listed in Tables 3 and 4 48 Switched patients followed 1 of the following algorithms: (a) those receiving a biosimilar for period 1 then switched to reference product, biosimilar, and finally reference product, which continued into extension, and (b) those receiving the reference product for period 1 then switched to biosimilar, reference product, and finally biosimilar, which continued into extension. Gerdes et al. (2017) pooled switched patients together to compare results against pooled nonswitched patients. 49 ■■ Discussion Some evidence from the comparison of biosimilar discontinuation rates supports the hypothesis that this outcome measure is highly susceptible to the nocebo effect. Median discontinuation rates of infliximab biosimilars for any reason, due to ADEs or lack of efficacy, were generally higher in open-label trials, suggesting that knowledge of a switch to a biosimilar product may have affected patient perceptions and subsequent outcomes. Etanercept biosimilar discontinuation rates for any reason were similar between study designs, but closer analysis of discontinuations due to ADEs also contradicts the theory of the nocebo effect. However, few studies were available evaluating a switch involving etanercept biosimilars. Clearer trends may be identified as additional studies with larger sample sizes are conducted.
Wide variability in the range of documented infusion and injection site reactions makes it difficult to draw a clear conclusion. In addition, the rate of etanercept biosimilar-related injection site reactions favors a trend opposing the nocebo effect. Ranges of objective ADEs, such as ADA development, were quite variable between study designs. Although hypothesized as unlikely to be susceptible to the nocebo effect, lack of reporting of baseline ADAs in double-blinded trials, coupled with relatively low reporting in open-label trials, prevented close evaluation of objective ADEs. Overall, the evidence is not sufficiently robust to clearly determine the presence of a nocebo effect during switches to a biosimilar product.
Analyses have reported biosimilars may represent up to $44 billion in potential cost savings over a 10-year period. 3,6 Unfortunately, uptake of biosimilars in the United States has been poor, in part due to prescriber discomfort with switching patients from reference products to biosimilars. 51 This review suggests that the negative expectations of patients may increase ADEs and discontinuation rates when switching to biosimilar products for infliximab and etanercept. Negative results from this phenomenon would then reinforce negative expectations of biosimilars and further hinder the possibility of biosimilar uptake in the United States.

Limitations
This review has some limitations that need to be considered. The overall number of randomized, double-blinded studies identified was quite small compared with the number of openlabel, observational studies, which makes it difficult to draw meaningful conclusions on the comparison. Comparing ADE rates across different studies may introduce many sources of bias. The patient populations in different studies varied greatly. The double-blinded trials did not include treatment of IBD, which was a large portion of the open-label trials. In addition, most data available were limited to one of the biosimilar products (CT-P13), but ADE rates may be product specific. The duration of follow-up was often short in open-label studies, which may decrease the ability to detect or report ADEs, especially those that may develop over prolonged exposure (e.g., ADA development). The frequency for monitoring of ADEs, especially ADA development, was higher in the double-blinded studies.
Many of the open-label studies did not report ADA monitoring practices or only conducted ADA testing after significant adverse reactions or loss of efficacy. In addition, the definitions for ADA development and confirmation of non-transient ADAs were not clearly defined in all studies. Finally, the heterogeneous population included in the studies may have introduced bias from an unknown or unevaluated source such as previous therapies, duration of previous therapies, or disease severity.

■■ Conclusions
Current evidence is insufficient to confirm a biosimilar nocebo effect, although higher discontinuation rates in infliximab biosimilar open-label studies support this theory. However, many limitations prevent drawing strong conclusions. Further studies are needed to evaluate the existence of a biosimilar nocebo effect. If it does indeed exist, the effects of mitigation strategies such as prescriber education and patient empowerment should be evaluated.
Strategies to combat the nocebo effect may be categorized as either conditioning or managing expectations of the patient and prescriber. Conditioning involves gradual introduction of the intervention or introduction of the intervention without patient knowledge. While gradual introduction may be technically feasible (e.g., dividing the total dose into part biosimilar and part reference product), this strategy is impractical and unlikely to be adopted by insurance providers. Switching patients to a biosimilar product without prescriber or patient knowledge presents several practical barriers, such as the lack of legality of biosimilar substitution without notification and the ability to identify the product prior to administration. This strategy may also be considered unethical and dishonest. [52][53][54] Thus, managing expectations through patient empowerment may be the most viable option for mitigating the biosimilar nocebo effect.
The most obvious way to manage patient expectations of biosimilars is through education. Many surveys have indicated that prescribers do not have a strong understanding of the manufacturing process, approval requirements, or ongoing regulation of biologic and biosimilar products. 55,56 This knowledge gap likely contributes to negative expectations. Prescriber hesitancy is probably a strong factor in creating or reinforcing patients' negative expectations and could enhance the nocebo effect. Appropriate framing of the discussion regarding switching to a biosimilar may help mitigate nocebo effects. An honest discussion should focus on the positive effects (cost savings), without intentional or subconscious hints that the biosimilar product is a "knock-off" product or possibly less safe or effective. Avoiding an overly focused discussion on potential ADEs has also been suggested to help manage expectations and decrease the nocebo effect. 52 The ethical implications of paternalistic nondisclosure to decrease nocebo-induced ADEs have been a controversial topic of debate that cannot be ignored. Transparency is an important principle in modern bioethics, raising the question of whether or not it is acceptable to waive our duty to inform patients in order to uphold nonmaleficence. 57,58 Regardless of which approach is taken, education for providers, pharmacists, and patients will certainly be an essential foundation to mitigating the biosimilar nocebo effect.