Answers are grounded in this paper and cite their sources.
“REQUIRED INFORMATION. An application submitted under this subsection shall include information demonstrating that” ‘‘(I) the biological product is biosimilar to a reference product based upon data derived from (aa) analytical studies that demonstrate that the biological product is highly similar to the reference product notwithstanding minor differences in clinically inactive components; (bb) animal studies (including the assessment of toxicity); and (cc) a clinical study or studies (including the assessment of immunogenicity and pharmacokinetics or pharmacodynamics) that are sufficient to demonstrate safety, purity, and potency in 1 or more appropriate conditions of use for which the reference product is licensed and intended to be used and for which licensure is sought for the biological product; (II) the biological product and reference product utilize the same mechanism or mechanisms of action for the condition or conditions of use prescribed, recommended, or suggested in the proposed labeling, but only to the extent the mechanism or mechanisms of action are known for the reference product; (III) the condition or conditions of use prescribed, recommended, or suggested in the labeling proposed for the biological product have been previously approved for the reference product; (IV) the route of administration, the dosage form, and the strength of the biological product are the same as those of the reference product; and (V) the facility in which the biological product is manufactured, processed, packed, or held meets standards designed to assure that the biological product continues to be safe, pure, and potent.’’ The BPCIA was amended at the end of 2022, with the term animal toxicology in Section (bb) being removed and replaced with nonclinical testing,1 which is described as follows: “cell-based assays, organ chips and micro-physiological systems, sophisticated computer modeling, other human biology-based test methods, and animal tests.” Animal toxicology testing represents one of the most irrational requirements; despite repeated reminders, such as for biosimilars approved in the United States and European Union, similar data packages were submitted to these regions. Despite the regulatory guidelines allowing waiver of animal studies based on analytical data, animal studies have been conducted for almost all the approved biosimilars in the United States and European Union.2 The new definition of “nonclinical” has long been suggested as pivotal in determining biosimilarity since these tests are more robust than clinical pharmacology and clinical efficacy testing.3 Clinical efficacy testing in patients is the primary contributor to the cost of biosimilar development; in contrast, clinical testing in healthy subjects, mostly to characterize PK, PD, and immunogenicity, contributes a tiny portion. It is not the FDA's responsibility to create guidelines for reducing development costs. However, when testing on patients is unnecessary, it falls within the purview of the FDA's responsibility to reduce unnecessary human exposure. After almost 2 decades of biosimilar use worldwide and the experience of regulatory agencies, there is a need to qualify these clinical efficacy testing studies.4 Accordingly, similar to other regulatory agencies, the FDA has identified alternative ways of evaluating patient efficacy. In January 2023, the FDA published a pivotal article reporting on clinical trials that support the waiving of clinical efficacy testing in patients if PD biomarkers show high similarity in clinical pharmacology studies conducted on healthy subjects.4,6 Considering that PD biomarkers are not available for most biological products (eg, monoclonal antibodies [mAbs]), the FDA has further suggested that developers should conduct intensive research using omics technology to identify such biomarkers. The FDA has also provided examples of this exercise, albeit not for mAbs. Another approach promoted by the FDA is using computational modeling to forecast adverse drug reactions caused by contact with biological receptors that are the targets of drugs with similar structures,7, 8 leading to the development of practical methods for predicting adverse events. Nevertheless, the criteria for surrogate biomarkers to support the approval of new drugs differ from those for PD markers to support a claim of biosimilarity.7 Furthermore, modeling approaches suitable for new molecules are unlikely to find any application in the establishment of biosimilarity owing to their complexity, uncertain understanding, and biosimilar developers’ limited ability to enter primary research areas. This study aimed to present the scientific contradictions in the current FDA guidelines for the approval of biosimilars that include clinical efficacy testing in patients, as well as the alternates to these studies proposed by the FDA. This study critically analyzes these alternatives to understand their application in developing biosimilars. This article presents a scientific analysis and concludes that clinical efficacy testing and PD biomarker comparisons, particularly when developers need to discover these, are much less sensitive for establishing biosimilarity than analytical assessment (including testing of receptor binding) and clinical pharmacology profiling with PK comparisons. The acquisition of ethics approval from an institutional review board or informed consent from participants does not apply to this article. The term clinical efficacy is incomplete without adding the qualifier “in healthy subjects” in PD marker comparisons or “in patients” in comparisons performed with clinical markers; this is how the FDA qualifies the trials. The sections below present scientific and logical arguments justifying why testing in patients should be waived because of its low sensitivity compared to other alternatives and the risk of qualifying products that may not have met the analytical and clinical pharmacology assessment based on clinical efficacy. As shown in Sections (cc) and (III), extrapolation of all indications of the reference product, including any future indications of the reference product, is allowed for biosimilars. The statute assumes that for a drug with multiple approved indications, testing for only 1 indication is sufficient, even if the various indications have different MOAs, as recently reported on antibodies, verifying that these differences cannot be identified from in vitro studies.8 For instance, the tumor necrosis factor-alpha (TNF-α) blocker adalimumab has at least 12 indications,9 with each indication having different MOAs and acting on different targets. Therefore, the allowance for all indications assumes that analytical assessment and clinical pharmacology profiling can establish clinical equivalence, as it would be irrational to test all indications; thus, efficacy testing is more sensitive and necessary.10 The extrapolation of indications to biosimilars11 assumes that the MOA for the biosimilar and its reference are the same according to an analytical match, irrespective of whether the MOA for the reference product is fully understood. This allows for the theoretical possibility that a biosimilar sponsor-developed PD biomarker may not be relevant, even if economically developing such a biomarker is possible. This is also a more robust exercise than a comparative PK study,12 and its validity remains questionable.13 The current understanding of the statistical modeling of clinical efficacy testing for biosimilars is based on study size calculation and, thus, study power. A new drug is tested against a placebo control when a significant difference is anticipated; however, a biosimilar candidate has already been proven to be the least different based on more objective analytical testing and clinical pharmacology profiling. A comparative trial must select an acceptable difference (equivalence margin) for study size calculation, at which point the rationality of biosimilar testing fails. An arbitrarily acceptable difference is selected in all instances because no data showing a dose-response relationship are available; this is labeled as “clinical judgment.” Despite the rationality of the selected equivalence margin, studies must meet the requirement for study power based on the FDA's requirement of a study power of 0.80 and a false positive rate of 5%. Given that a biosimilar candidate has already been shown in analytical and clinical pharmacology comparisons to be highly similar, the calculated study size will always be higher than the study size when a new product is tested against a placebo. As shown by a recent study, if the significance level of actual proportions is set at 0.28 and 0.33, more than 26,000 subjects will be required to demonstrate a difference, with an equivalence margin of 0.06.14, 15 Notably, such a challenge in designing clinical trials has never been faced before. A good analogy is to divide the same batch of a product into 2 parts and subsequently determine whether a difference exists between the 2. The inherent inter- and intrasubject variability in the testing of biological drugs, including those used in cancer treatment, further complicates the uncertainty of study power. In such cases, listing naïve patients is impossible, many of whom do not survive throughout the study course. In summary, studies with an arbitrarily acceptable difference in clinical response, uncertainty in response measurement (even if the response is a checkpoint), high variability, or patient variability cannot add value to establishing biosimilarity. The Bayes theorem can be used to determine the accuracy of clinical efficacy testing results by considering how likely any given study is accurate. Prior probability is the probability of an event occurring before new data are collected. In other words, it represents the best rational assessment of the probability of a particular outcome based on current knowledge before an experiment is performed. Therefore, the Bayes theorem of conditional probability can be restated as: Posterior = Likelihood * Prior/Evidence The study design defines the likelihood and evidence as required by the FDA, leaving the posterior probability dependent on prior probability or how often such studies have failed. Several studies have drawn the conclusion mentioned above. For instance, a recent meta-analysis of clinical testing reports on oncology biosimilars showed that 31 studies involving more than 12,000 patients met the biosimilarity criterion, with only one demonstrating a slightly higher efficacy.16 In 2 studies showing immunogenic response, evaluation was omitted, which is impossible nowadays. Additionally, the European Medicines Agency (EMA) or FDA requested no withdrawals or recalls from the market owing to the failure of clinical safety/efficacy testing in patients. Furthermore, the clinical trials website data indicated that none of the 141 reported studies failed, and comparative efficacy testing did not lead to product withdrawal or recall from the market. These data are available for 108 European Public Assessment Report files from the EMA and 44 approval documents from the FDA. No failed studies were reported when conducted in the comparator mode, and the acceptance criteria in some cases were retroactively revised to meet the equivalence criteria. Over 200 studies with reported results in the ClinicalTrials.gov portal have satisfied the acceptance criteria, and the PubMed database lists 504 randomized controlled clinical trials showing no clinically meaningful differences. These conclusions have been reported for years. For example, a review published in 2017 17 concluded that 10 comparison studies in the cynomolgus monkey across 4 monoclonal (mAb) classes showed similar exposure in all cases. While animal toxicology studies are no longer required, in the past, in 17 toxicology comparison studies with 5 mAb classes performed in the same species and 7 toxicology comparison studies with non-mAb biosimilars in the rat, no new/unexpected findings were seen, and drug exposure measurement gave comparable values in all cases. Recently,18 detailed analyses of the marketing authorizing applications of all 33 mAbs and 3 fusion proteins were evaluated by the EMA between July 2012 and November 2022. In 67% of cases, the quality and clinical assessment outcome was the same, that is, both the quality and clinical assessments either supported or did not support approval. In 11% of cases, significant objections were identified in the quality part of the submission but not in the clinical data. In 22% of cases, major objections were raised about the clinical data package but not the quality data. However, in no instance did seemingly negative clinical data, including failed efficacy trials, lead to a negative overall decision. In each instance, the failure to confirm similar clinical performance in all investigated aspects was eventually viewed as not being related to the biosimilar per se but as being due to imbalances in the trial arms, immaturity of secondary end point results, change in the reference product, or even chance findings. In no case were clinical trial data necessary to resolve residual uncertainties regarding the quality part. The conclusions further support the argument that sufficient evidence for biosimilarity can be obtained from a combination of analytical and functional testing and PK studies, which may also generate immunogenicity data. This calls into question the usefulness of comparative efficacy studies for regulatory decision making when approving biosimilar mAbs and fusion proteins.41 Clinical efficacy testing of new drugs against a placebo is considered the gold standard. Nonetheless, this has recently been questioned for real-time testing, with claims that clinical efficacy testing is “broken”19 and that new digital technologies and real-world evidence are necessary given the 21st Century Cures Act.20 While an analytical match between a biosimilar and its reference is essential and the analytics upon which that match is established continue to improve,21 the sensitivity of clinical studies remains limited,22 and what is meaningful can be a matter of judgment. Hence, primary and secondary end points for originator product development and pivotal clinical studies on each licensed indication are important. The biosimilar does not need to provide clinical data on all reference indications because extrapolation between the biosimilar and everything already established clinically for reference can be carried out.11 Regulatory bodies are aware of the limitations of efficacy testing in patients. Current alternatives to efficacy testing include comparisons between the PD properties of a biosimilar candidate and those of its reference product, as suggested by the FDA's Division of Applied Regulatory Science.2 The FDA recommendations do not include MOA biomarkers; instead, the FDA suggests the application of omics technology (eg, metabolomics, transcriptomics, and other “omics”) to discover new chemical biomarkers (including the proteins formed by the drug).35 The FDA has also validated that PD biomarkers could be identified using large-scale proteomic approaches and other technologies; however, studies by the FDA did not include testing for mAbs. Finding a new PD biomarker involves extensive characterization of blood samples using high-end mass spectrometry and other more sophisticated technologies to identify a chemical (such as a protein) that appears only in subjects receiving the biosimilar candidate and the reference product but not in the control group. A good analogy of this exercise is, more colloquially, “finding a needle in the haystack” not knowing what the needle looks like. The complexity of this scientific research involves establishing validated test methods without correlating the levels with clinical response. This allowance by the FDA makes this testing of little value because any finding of a difference or lack of difference cannot be translated into a claim of efficacy. There is no way to investigate whether a nonlinearity exists in the concentration and its impact on the clinical response. These studies are unnecessary or offer no assurance of biosimilarity, compared to what has already been established from PK and PD data. The FDA should allow biosimilar developers to select practical markers instead of developing new biomarkers; nonetheless, there should ideally be no PD biomarker comparisons. The scientific and economic challenges posed by suggestions to introduce new biomarkers have been questioned because more robust and relevant biomarkers, including receptor-binding comparisons, are already available. However, understanding the MOAs through more relevant in vitro functional assays can facilitate the more efficient development of biosimilars. Considering that receptor binding initiates a PD response, cell-based bioassays, or potency assays, such as enzyme-linked immunosorbent assays, binding assays, competitive assays, cell-signaling assays, ligand-binding assays, proliferation assays, and proliferation suppression assays, should enable a sufficient comparison between the functional properties of a biosimilar candidate and those of its reference product. Additionally, while typically not necessary, functional tests for MOAs, such as apoptosis, complement-dependent cytotoxicity (CDC), antibody-dependent cellular phagocytosis, and antibody-dependent cellular cytotoxicity, can be added to boost confidence in the safety and efficacy of biosimilars. A therapeutic response occurs when mAbs attach to the protein epitope targets in target and objective methods for establishing functional similarity include the use of target for binding to particular receptors and such as antibody-dependent cytotoxicity and such as and and binding using in or in vitro such as enzyme-linked immunosorbent testing for of regulatory of tumor necrosis target suppression of and as a blocker may also be assays are more robust than testing in patients in of establishing efficacy comparisons without demonstrating any PD response for However, functional tests cytotoxicity, antibody-dependent cellular phagocytosis, and have little value when a drug targets a A of functional assays to a of biological can be for products with multiple biological proteins and receptor-binding across functional is a of biological and is established using various The technology for testing has the Several methods for evaluating receptor binding and other quality of biosimilar with their reference products are allowing a approach to establishing a PD response is in a of the protein to its a The clinical response to biological drugs on their which with receptor Current has this test highly and Additionally, mAbs can with multiple receptors and be evaluated through analytical The primary receptors in the of therapeutic proteins are as within the of such 1 receptors receptors receptors 2 receptor 2 receptor receptor receptor and that receptor-binding tests are more sensitive to PD and clinical response tests than functional or biomarkers. An approach for establishing comparable receptor binding would be more robust and objective than and testing an PD marker to establish biosimilarity. efficacy testing, acceptable differences are always based on clinical many clinical pharmacology studies PK, PD, and immunogenic using acceptance criteria. As all and their on the PK a highly comparable PK should support PD The FDA has also promoted in PK studies instead of testing in patients and has suggested that immunogenicity studies may not be necessary if immunogenicity does not the PK As in the FDA's guidelines on therapeutic investigated in the target because animal testing and in vitro cannot the human response, immunogenicity assessment a in demonstrating product and in the of biosimilar minor can or safety, including the immunogenicity of a therapeutic Several in vitro immunogenicity assays the risk of exposure in healthy subjects and can be used in clinical pharmacology profiling. The argument that patient testing is necessary for safety evaluation is because the same testing in clinical pharmacology profiling in healthy subjects is more given that patients are unlikely to have systems, allowing for a In to analytical a comparative PK study between a biosimilar and its reference originator product is the most sensitive for differences. biosimilars in patients has been to not add confidence in safety and efficacy demonstrating high similarity in analytical assessment and clinical pharmacology profiling. such as allowing extrapolation under the of MOA an arbitrarily acceptable difference, the study size and the application of the Bayes such testing less than the testing already conducted in analytical assessment and clinical pharmacology profiling. The FDA's suggestions to clinical pharmacology testing by and adding PD biomarkers are practical given the scientific and economic of these suggestions are necessary because already have more robust receptor binding and PK profiling testing to establish biosimilarity. efficacy testing as a exercise as in unnecessary human testing should be from justifying analytical and differences based on patient efficacy While testing in patients is to be the of safety and a more understanding that the compared products have already been proven to be highly similar makes such testing less sensitive and a that is to even in the of regulatory at one the of efficacy testing at the other the of even less relevant and more The scientific arguments in this article are to about a major change in the regulatory approval for biosimilars. This will the for which biosimilars were to about more but not at the cost of safety and efficacy. The extensive use of to support in originator their has established regulatory confidence in using analytics and quality to clinical In when PK is compared and or between a biosimilar and its the power to differences is It is not the PK per se that is the most but the ability to tiny differences between 2 test products in a the sensitivity of a PK study for showing a match in the response to a biological is even if the clinical of that response from the experience of the originator product the change a PD biomarker can further the PK match but does not add In summary, clinical efficacy testing and PD biomarker comparisons, particularly when developers need to discover these, are much less sensitive for establishing biosimilarity than analytical assessment (including testing of receptor binding) and clinical pharmacology profiling with PK comparisons. The no of This study did not any The is not for the or of any information by the than should be to the for the article.
No takes yet. Share an insight, caveat, or question.
Sarfaraz K. Niazi (2023) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: