“Development of Fixed Dose Combination Products” Workshop Report: Considerations of Gastrointestinal Physiology and Overall Development Strategy

The gastrointestinal (GI) tract is one of the most popular and used routes of drug product administration due to the convenience for better patient compliance and reduced costs to the patient compared to other routes. However, its complex nature poses a great challenge for formulation scientists when developing more complex dosage forms such as those combining two or more drugs. Fixed dose combination (FDC) products are two or more single active ingredients combined in a single dosage form. This formulation strategy represents a novel formulation which is as safe and effective compared to every mono-product separately. A complex drug product, to be dosed through a complex route, requires judicious considerations for formulation development. Additionally, it represents a challenge from a regulatory perspective at the time of demonstrating bioequivalence (BE) for generic versions of such drug products. This report gives the reader a summary of a 2-day short course that took place on the third and fourth of November at the Annual Association of Pharmaceutical Scientists (AAPS) meeting in 2018 at Washington, D.C. This manuscript will offer a comprehensive view of the most influential aspects of the GI physiology on the absorption of drugs and current techniques to help understand the fate of orally ingested drug products in the complex environment represented by the GI tract. Through case studies on FDC product development and regulatory issues, this manuscript will provide a great opportunity for readers to explore avenues for successfully developing FDC products and their generic versions.

the present motility [14]. In the seventies, Vantrappen et al. observed a higher secretion rate of bicarbonate 132 shortly after an upper GI phase III contraction [3]. In doing so, the gastric acid of the stomach entering the 133 small intestine could directly be neutralized by the bicarbonate buffer. This so-called 'secretomotor 134 complex' is highly likely to be a responsible factor in the formation of water pockets inside the intestinal 135 tract. Besides gaining knowledge with respect to the present volumes in the GI tract, the composition of 136 these fluids is another important aspect. In a recent study, human duodenal fluids were aspirated from 20 137 healthy subjects in the fasted and fed state [18]. The fed state was simulated by ingestion of a liquid meal 138 (i.e., 400 mL of Ensure Plus®, equal to 700 calories). After aspiration of these fluids as a function of time, 139 fluids were analyzed for pH and endogenous constituents (bile salts, phospholipids, cholesterol, enzyme 140 activity and lipid digestion products). The results of this study demonstrated wide variability in the presence 141 of these constituents from person to person, although the study protocol was the same for each and every

160
Besides solubility, absorption has always been a key parameter in estimation of drug performance. 161 Multiple techniques are described in the literature to assess the intestinal permeability of drug compounds. 162 The Loc-I-Gut ® method, i.e., a double-balloon perfusion system, is an interesting study technique to explore 163 the permeability for drug compounds in the different regions of the GI tract [20,21]. A specific region of 164 the GI tract will be inflated by two balloons and thus separating a specific region of interest. Subsequently, 165 a drug solution will be perfused and the amount of drug that will disappear is a measure for the amount of contractions, including the high-amplitude propagating contractions associated with movements of solid 178 colon content, represent a minority of the colonic activity and are normally more frequent about 1-2 hours 179 after the meal and upon awakening [24]. The reason of this is likely related to the fact that, in these moments 180 of the day, the arrival of the content accumulated in the distal small bowel during the night and during the 181 inter-digestive periods determine the distension of the ascending and transverse colon that trigger the 182 propagating activity. The prevalence of non-propagating and retrograde activity explains the fact that the 183 normal colonic transit time is slower (about 35 hours) as compared to the small bowel. This allows the 184 colon to perform its functions of absorption, fermentation and to be an adequate reservoir organ. HRM is a 185 useful technique to study colonic motor function but is invasive and normally requires a preparation of the 186 bowel. This makes the technique less attractive when the colonic function needs to be studied under 187 physiological conditions. Recently, other techniques have been applied to study the colonic function. The The GI tract is a complex and not well-understood sequence of organs with changing environments 204 as a function of time. However, an in-depth mechanistic understanding of the obstacles and opportunities 205 in each segment is necessary to achieve optimal drug absorption and bioavailability (BA) (Figure 2).   Motility effects and gastric emptying are know to have an impact on the performance of a drug product but 214 they are seldom considered in drug development. In contrast, food effects, pH effects and solubilization 215 effects by bile salts were studied intensively in the past decades. However, today, there is still no consensus 216 prediction error (%) regarding simulated plasma Cmax and AUC were7 and 14% , respectively, when using 223 these biorelevant solubility value as an input in GastroPlus™. Solubility values obtained in aqueous media 224 (pH 6.5) resulted in a 38 and 63% prediction error with respect to plasma Cmax and AUC.Later on, a dynamic 225 dissolution protocol was developed in biorelevant media (i.e., FaSSIF) which again showed predictive 226 power for establishing an IVIVC [30]. The dynamic dissolution protocol was then applied to a flow-through 227 apparatus for montelukast sodium. Again, the biorelevant media gave the best fit to clinically observed data 228 [31]. These early studies were successful to establish IVIVC without considering other GI factors. In a 229 study by Almukainzi et al., the impact of gastric motility on the pharmacokinetics (PK) of meloxicam was 230 studied [32]. It was observed that two formulations (conventional versus fast dissolving) had a similar PK 231 pattern when administered in a rodent model. However, when the gastric motility was impaired the stomach 232 controlled the drug release and therefore the drug absorption for the conventional dosage form. The PK of 233 the fast dissolving formulation was close to the pattern observed in the healthy state. This study indicated 234 that formulation differences, which are not relevant under healthy conditions, might result in significant 235 differences under disease state. This study showed that the stomach in disease conditions is able to 236 negatively impact PK parameters such as plasma Cmax and Tmax. Furthermore, it is well accepted that gastric 237 emptying impacts the PK in fasted versus fed state for many drugs. However, less attention is given to the 238 fact that GI motility impacts Cmax and Tmax depending on the dosing time and the MMC phase. This might 239 be due to the fact that the PK models used to quantify and describe the PK behavior of drugs sooth-out 240 individually observed variability in the mean PK profiles. However, if motility and PK are both monitored 241 a relationship between observed plasma levels and intestinal motility are getting more obvious. Another 242 factor for alternations in drug absorption is the composition of the intestinal juices. The buffer system in 243 the GI tract is carbonate-based. In routine pharmaceutical quality control (QC) and development, phosphate 244 buffers play a major role while carbonate buffers are seldom used. The choice of phosphate over bicarbonate 245 seems to impact the in vivo performance of enteric-coated dosage forms. Early reports show the failure of 246 enteric-coated products in vivo (1964) and are confirmed over several decades until today by in vivo studies 247 This drug is absorbed to over 80% in 2 hours but it takes about 15-20 hours to observe the maximum 265 fraction dose absorbed. A classical IVIVC would correlate fraction dose absorbed versus the dose dissolved. 266 However, in this specific case, the IVIVC would be misleading. The drug dissolves fast in the gut and is 267 completely dissolved within 15 min. As mentioned before, >80% will be absorbed into the enterocytes 268 within 2 hours. The drug undergoes lysosomal trapping after entering the enterocyte. As a weak base, it is 269 highly lipophilic at physiological pH in the cytoplasm. As the drug will migrate through the enterocytes 270 from the apical to the basolateral side, it can pass through the membranes of the lysosomes and it can enter 271 into an aqueous environment with a slightly acidic pH. In this organel, the weak base becomes more 272 hydrophilic and, therefore, will be entrapped in the lysosomes. That isis why it takes more time to appear 273 in the blood than it takes time to be absorbed. Such drugs dissolution tests are not useful surrogates for in 274 vivo performance since the dissolution of the drug product cannot be directly correlated to the plasma levels. 275 It is the biological system and its specific environments and drug partition between the cell compartments, 276 that determine the appearance of the drug in the central compartment and not the drug dissolution. In 277 summary, GI drug absorption is highly impacted by different physiological factors. In vitro performance 278 testing should consider and include physiologically-adapted test protocols to identify potential clinical 279 relevant dosage form factors. A BCS sub-classification system, which includes acids, bases and neutral 280 molecules can help to identify potential obstacles for oral drug absorption for these different groups [39]. 281 To meet all these standards, a potential in vitro apparatus, which can simulate the different GI conditions, 282 is shown in Figure 3.  In the first part of the presentation, the relevance of exploring the biopharmaceutical properties of each drug 293 in the combination product were discussed in the framework of different classification systems. The BCS 294 system has evolved from a regulatory conservative classification framework in which the main concern is 295 to ascertain the non-bioequivalence (non-BE) risk to a development tool which can help on the formulation 296 strategy selection [40,41]. In order to understand the biopharmaceutical limiting factors for a given drug were sub-classified in neutral (BCS IIc), weak acids (BCS IIa) and weak bases (BCS IIb). Following these 306 sub-divisions, the suggested dissolution tests to forecast in vivo behavior differ from class I and III for 307 which simple dissolution apparatus (as USP II) could suffice and from class II and IV for which a gastric 308 compartment and an absorptive sink should be included in order to increase the in vivo predictability. To 309 accommodate that need, several dissolution system have been proposed in the literature and as example 310 several transfer systems and two-phase or biphasic dissolution systems were described [37,42-47]. In the 311 second part of the lecture, the potential effects of formulation excipients were discussed in as well as 312 experimental preclinical models to study those effects. Excipients can affect membrane permeability and 313 metabolism and GI motility either at gastric emptying level or at intestinal level. In Table 1   preclinical models is necessary to assess formulation performance. 337

Challenges and Opportunities to Grant BCS and Dose Strength Based Biowaivers for FDC 338
Products BE studies are required in order to bridge pivotal clinical data of the reference listed drug (RLD) 365 product(s) to safety and efficacy of FDC products belonging to Scenarios I and II. While the design of BE 366 study for scenario I is standard, in scenario II the in vivo performance (e.g., PK end-points) of the FDC 367 product is compared to the co-administration of the SEPs. In both cases, successful BE indicates the absence 368 of (or similar) PK interactions between APIs. However, BE studies for FDC products are challenging due 369 to: i) potential changes in PK intra-subject variability in the combination product; ii) non-linear PK in a line 370 of strengths; iii) drug-formulation interactions; and iv) differential impact of food on API PK when 371 there is a consensus that DDI or DFI might be of minor clinical relevance for BCS class I drugs, there also 412 a concern that these interactions could greatly impact the oral absorption of low permeability APIs. FDC 413 products also offer opportunities for developing a line of strengths that can be used to optimize therapy by 414 dose titration. Intermediate and low strengths could apply for a dose strength (DS)-based biowaiver 415 provided there is at least one strength (typically the highest) that successfully demonstrated BE to the 416 reference product in vivo. Dose strength-based biowaivers are applicable to APIs that are not eligible for 417 BCS-based biowaivers and to pharmaceutical forms other than IR (i.e., modified-release, delayed-release). fulfilling compositional requirements for FDC products based on segregation technologies (e.g., bi-layer 428 tablets) since EMA treats each layer as a separate entity while FDA considers bi-layer tablets as a single 429 unit. Also, in the case of single unit FDC products with large dose disparity between APIs it might be very 430 difficult to fulfill proportionality requirements by both FDA and EMA. More specifically, EMA states that 431 in order to calculate API/excipients proportionality the other API must be considered an excipient. 432

However, it is not clear whether the other API must be considered as a filler for proportionality calculations. 433
Similarly, there is no specific FDA recommendation as to how to consider the other API in bi-layer tables. 434 These discrepancies can hinder simultaneous registration of an FDC product in both USA and Europe.    However, underpowered studies with too many variables can further confound an already complex issue 468 and should be avoided. Pivotal BE studies should be designed with due consideration of all the 469 physicochemical, biopharmaceutic, and PK data for the compound from all sources. BE study designs 470 specific to highly variable drugs such as scaled BE or cross-over replicate designs may be considered. 471 Leveraging the knowledge gained from varying but synergistic techniques such as in vitro 472 solubility/dissolution studies, in silico absorption models and IVIVC's, in vivo preclinical animal models, 473 and the available in vivo clinical data is paramount to the success of the FDC strategy for a given 474 combination. Two case studies were discussed where the use of oral absorption modeling, dissolution data 475 and clinical PK data were used to successfully develop FDC products. In the first case study, the 476 development of a triple combination product was discussed, where one of the active ingredients had a 477 highly variable Cmax and another active had a long Tmax due to bile secretion and slow absorption. In this 478 case oral absorption modeling was key to understanding the impact of formulation changes on PK of the 479 three actives and ultimately in development of the FDC product. In the second case study, the development 480 of a double combination product was discussed, where one of the active ingredients was a weak base with 481 high intra-subject CV and steep pH-solubility profile. In this case, data from several relative BA studies 482 and a thorough understanding of the PK and biopharmaceutic properties helped with the successful 483 development of the FDC. 484

(EU/USA/ Latin America/Japan) -Alexis Aceituno, PhD 486
Although one of the purposes is to combine drugs at fixed dose ratios to simplify treatment of 487 chronic diseases and improve patient adherence, there is a general consensus that this rationale cannot be there is a deficient response to one or more drugs to be included in the proposed combination. Drug-Drug 503 (DDI) or PK interaction study may be required if the combination poses a threat with potential clinical 504 consequences; 2) substitution by an FDC product when a reduction of pill burden is sought after. 505 Bioequivalence testing is required and special attention should be paid if the FDC product is dosed at 506 different time intervals, and 3) FDC therapy initiation if the FDC product has not been used previously for 507 any particular indication. Both clinical and pk trial, as well as DDI study, should be performed and 508 submitted prior to approval. In Latin America, there is only one specific guidance for registration of FDC 509 products since 2010 [84]. It describes the definition of FDC products, general consideration for filing and 510 regulatory requirements that depend on the proposed dose scheme or the drugs to be combined. FDC 511 approval can be granted under the following conditions: 1) An FDC product contains the same actives, dose 512 and dose regimes as mono products used concomitantly, therefore the safety and efficacy profiles are well 513 known; to demonstrate efficacy, a bioequivalence study may be sufficient; 2) same conditions as in "1", 514 but FDC product is going to be used in novel dose or new therapeutic indication and therefore a phase III 515 clinical trial is required; 3) the combination contains one or more new active ingredients and phase I, II and 516 III clinical trials are required to gain approval. In general, there is not a globally applicable guideline for 517 FDC product registration, but for specific therapeutic classes and four general cases that are described in a 518 WHO technical report, aiming at guiding pharmaceutical companies for development, approval, and 519 marketing FDC products under less developed jurisdictions [61]. Although generic and hybrid submission 520 pathways seem to be sufficient under most jurisdictions, preclinical and clinical data for novel combinations 521 will always be needed if individual components in FDC products are either known or they are new 522 investigational drugs. However, the idea still persists among regulated entities that different jurisdictions 523 around the world should give more importance to convenience/compliance as a rationale for developing 524 FDC products either containing authorized/new drug entities or authorized drugs only bearing in mind 525 patient´s satisfaction or reduced/contained health costs [85]. If generic development is allowed, a BE study 526 design for a FDC product should consider the same principles as if the drugs were given alone, looking for 527 the achievement of equivalence in PK profiles for each FDC active ingredient and their respective either 528 reference FDC or reference mono products. At this point, it is important to realize that PK interactions may 529 have more critical consequences with FDC products than the same drugs given as mono products 530 concomitantly. To conclude, when comparing jurisdictions to obtain FDC product approval, it seems 531 necessary that a balance should be reached between an overcautious registration approach and the potential 532 large public health benefits that would arise from affordable FDC products of proved efficacy. The 533 achievement of broad harmonization in the understanding and application of existent technical guidelines 534 and requirements for FDC product development and registration is still a pending matter. 535

(FDC) of Oral Solid Dosage Forms -Divyakant Desai, PhD 537
For formulation scientists without prior experience of the FDC development, two decision trees 538 were discussed to select the most suitable formulation development strategy. The first decision tree was 539 related to the formulation design for an FDC product (Figure 4). If two drugs are chemically incompatible, multi-layer tablet or a drug-specific multi-particulate 556 system was proposed. If they are compatible, then a monolithic system was proposed unless there is a need 557 to keep them apart in order to maintain the dissolution profiles comparable to the respective single entity 558 product. The second decision tree was about the selection of the manufacturing process for an FDC product 559 ( Figure 5).  The drug loading in the formulation dictated the selection of the manufacturing process. If the drug 576 loading is high, a hot melt extrusion (HME) or a bi-layer method of manufacturing was proposed. For a 577 formulation with a low drug loading, an active coating approach was proposed. One of the crucial factors 578 in the manufacturing process selection is a pharmaceutical scientist prior experience with the manufacturing 579 process under consideration. A monolithic formulation system, where two drugs are incorporated in a single 580 dose unit, is considered the most simple formulation approach. However, a case study was presented where 581 a second drug, hydrochlorothiazide (HCTZ), was added to the existing formulation of a hypertensive drug 582 [87]. It was shown that povidone (a binder) and poloxamer (a wetting agent) triggered HCTZ degradation 583 under accelerated storage conditions by solubilizing HCTZ in available moisture. Replacement of povidone 584 by Starch 1500, resolved the stability issue and removal of poloxamer did not impact the BE study 585 adversely. For a bi-layer tablet formulation approach, which is normally used to keep two incompatible 586 drugs apart or to maintain two drug release profiles, few critical formulation factors were presented. Those 587 factors include the selection of excipient with high fragmentation tendency such as lactose in the first layer, 588 more deformable material such as microcrystalline cellulose in the second layer, the weight ratio of not 589 more than 1:6 for two layers. It was also emphasized that the tamping force for the first layer should be able 590 to reduce the volume without sacrificing the surface roughness which is essential for the adhesion of the 591 second layer. Two case studies were presented on the bi-layer formulation approach. In the first case study, 592 the compressibility of an extended release metformin formulation was improved by the addition of 1% w/w 593 silicon dioxide. In the second case study, two different grades of fumed silica behaved differently in a bi-594 layer tablet formulation [88]. Aerosil 200 did not cause layer separation but Aeroperl 300 did. Aeroperl can 595 adsorb relatively large amounts of moisture at any humidity level due to its greater surface area, but it does 596 not retain moisture when the humidity decreases. In contrast, Aerosil adsorbs relatively smaller amounts of 597 moisture but it retains moisture due to its large pore sizes. It was hypothesized that the moisture not retained 598 by Aeroperl could be available for interactions with other layer excipients such crospovidone. The third 599 formulation technique presented was an active coating technology. An active coating can also be used to 600 maintain two separate release profiles and to separate two incompatible drugs. A case study was presented 601 to show how acid and base sensitive molecule was stabilized selecting and minimizing the excipients in a 602 coating material API come in intimate contact with. For example, 1 mg drug is placed with 99 mg of 603 excipients for a 100 mg tablet, the 1 mg drug can react with 99 mg of excipients. However, if 1 mg drug is 604 placed with 9 mg of coating material, the amount of available for a reaction is reduced drastically. It is also 605 a useful technology to make a tablet for a compression sensitive molecule. Although the active coating is 606 useful, it is not as widely used as other technologies because it presents two big challenges. The first 607 challenge is how to detect coating endpoint so that tablets with correct potencies can be manufactured. If a 608 coating process is stopped early, tablets may be sub-potent. On the other hand, if the coating is stopped 609 late, tablets may be super potent. The second challenge is content uniformity (active coat uniformity). The 610 content uniformity can be influenced by various process parameters such as pan load, coating time, number 611 of coating guns, and spray quality. A mathematical model was presented in which model parameters were 612 linked with the process parameters for scale-up. It was shown that the model correctly predicted coating 613 uniformity of tablet weighing 200 mg to 1450 mg in different shapes at a 450 kg commercial scale. In 614 summary, the decision trees are very useful to explore the most suitable formulation and manufacturing 615 process for an FDC formulation. Each formulation approach for an FDC will have its own unique challenges 616 but as illustrated by various case studies, it is possible to overcome these challenges to develop a rugged 617 formulation and a commercially viable manufacturing process using various process analytical technologies 618 (PAT). 619

Clinical Pharmacology Aspects of Fixed-Dose Combination Drug Development -Dakshina Murthy 620
Chilukuri, PhD 621 Combination products are defined in the Code of Federal Regulations [21 CFR 3.2 (e)]as categories 622 of drug-drug combination products. These products could be two or more approved drugs or investigational 623 drug(s) developed along with an approved drug(s) or two or more investigational drugs developed together. 624 The final products can be FDCs, co-packaged products or separate individual products administered 625 together. Among the reasons why these products are developed are the additive/synergistic effects of drugs 626 for the same disease (e.g., anti-viral and cough/cold drug products). Sometimes when two drugs have

Concluding Remarks and Future Perspectives 660
Market access for FDC products is challenging in terms of achieving bioequivalence to co-661 administration of the individual mono-products, but also because of formulation challenges (compatibility 662 of API's, doses). However, we should not neglect the impact of GI physiology on oral drug behavior which 663 can result in intersubject differences in systemic outcome, potentially leading to failures in bioequivalence 664 studies. Therefore, it's important to finalize a clear link between formulation strategy and clinical 665 evaluation, supported by guidelines of regulatory authorities. In addition, the contribution of in vitro 666 predictive dissolution testing can help assist regulatory decisions with respect to the approval of FDC 667 products in a sense that these models identify the underlying GI variables playing a crucial role in the 668 absorption process inside the GI tract. From an academic point of view, these clinically-relevant dissolution 669 models can be optimized and validated when pharmaceutical companies would share their non-BE 670 formulations (i.e., clinical failures). When they do so, the underlying problems can be unraveled which will 671 be taken into account by formulations scientists when formulating FDC products. This report represents the scientific views of the authors and not necessarily that of the regulatory 677 authorities presented in this manuscript (U.S. Food and Drug Administration and ANAMED). 678