key: cord-0861604-4ek9prsg authors: Nugrahani, Ilma; Parwati, Rismaya Desti title: Challenges and Progress in Nonsteroidal Anti-Inflammatory Drugs Co-Crystal Development date: 2021-07-09 journal: Molecules DOI: 10.3390/molecules26144185 sha: 3f37da2fed3ae5abdac977e26e0eb3006a606a98 doc_id: 861604 cord_uid: 4ek9prsg Co-crystal innovation is an opportunity in drug development for both scientists and industry. In line with the “green pharmacy” concept for obtaining safer methods and advanced pharmaceutical products, co-crystallization is one of the most promising approaches to find novel patent drugs, including non-steroidal anti-inflammatory drugs (NSAID). This kind of multi-component system improves previously poor physicochemical and mechanical properties through non-covalent interactions. Practically, there are many challenges to find commercially viable co-crystal drugs. The difficulty in selecting co-formers becomes the primary problem, followed by unexpected results, such as decreased solubility and dissolution, spring and parachute effect, microenvironment pH effects, changes in instability, and polymorphisms, which can occur during the co-crystal development. However, over time, NSAID co-crystals have been continuously updated regarding co-formers selection and methods development. Non-steroidal anti-inflammatory drugs (NSAID) are antipyretic, anti-inflammatory, and analgesic agents commonly used to treat muscle pain, dysmenorrhea, rheumatism, pyrexia, gout, migraine, etc. These drugs also are combined with opioids to treat acute trauma cases [1] . The use of NSAID is increasing every year. In 2018, NSAID prescriptions increased by 40.7%, the most significant percentage prescribed to women on long-term treatment. Paracetamol and ibuprofen are the most frequently prescribed NSAID for chronic therapy using fixed-dose combination treatment [2] [3] [4] . Based on their chemical structure and selectivity, NSAID are divided into salicylic acid derivatives (aspirin, ethenzamide, diflunisal), para-aminophenol derivatives (paracetamol), indoles (etodolac, indomethacin), hetero-acyl acetic acid (diclofenac, ethyl diclofenac, ketorolac), aryl-propionic acid derivatives (ketoprofen, naproxen, ibuprofen, flurbiprofen), anthralinic acid derivatives (mefenamic acid, flufenamic acid, nifluminic acid, meclofenamic acid), enolic acid derivatives (tenoxicam, piroxicam, meloxicam), and diarylheterocycles or selective cyclooxygenase (COX)-2 inhibitors (celecoxib, etoricoxib) [1] . NSAID are classified in the Biopharmaceutics Classification System (BCS) as class II, because they have low solubility and high permeability [5, 6] . The low solubility of NSAID can cause many problems in using and producing these drugs, especially regarding the onset of action. NSAID are the most widely used drugs for pain relief, so a fast onset of action is needed [6] . It is a challenge to develop an optimal potential effect. Various attempts have been made to increase the solubility, such as the formation of salts, polymorphs, solvates, and hydrates [7] . Nowadays, most NSAID on the market are in their salt form, such as diclofenac sodium, diclofenac potassium, ibuprofen, etc. However, salt formation can only be applied to active pharmaceutical ingredients (APIs), forming ions. It is a problem for some compounds that are difficult to ionize, such as zaltoprofen (ZFN). ZFN is included in the propionic acid group and is an effective NSAID to treat acute somatic melting point of the co-crystal, which differs from those of the parent drugs [31, 33] . The differences in diffraction peaks and intensity in PXRD inform the new crystal interaction, i.e., interactions between the oxygen atoms (O) and the H atoms of primary amide groups, which SCXRD can structurally determine [31, 34] . A change in the solid form indicates crystalline formation, which SEM can characterize [34] . Computational studies have shown co-crystal interactions through the calculation of free energy using Schrodinger software. For example, piroxicam (PXC)-sodium acetate (SAT) co-crystal was reported to have lower free energy (−1671.29 kJ) compared to PXC (−1442.71 kJ) and SAT (−228, 49 kJ). Low free energy indicates that the co-crystal is stable [33] . NSAID co-crystals were first reported and patented in 1973 by Gerhard Dorler and Maria Kuhnert-Brandstetter. Over the decades, NSAID have been developed into cocrystals, salt co-crystals, co-crystal hydrates, salt hydrate co-crystals, and drug-drug co-crystals. For example, a co-crystal consisting of two APIs, namely pyrithyldione as sedative and propyphenazone as NSAID, produces three polymorphs; the polymorph form II was the most stable [35] . Since then, NSAID co-crystals have continued to develop in terms of screening, manufacturing techniques, and enhancement of physicochemical properties [23, 28, [36] [37] [38] . Furthermore, some invented NSAID co-crystals are summarized in Table 1 . Table 1 . NSAID co-crystal development. Ethenzamide-gentisinic acid 1:1 Solvent evaporation Increased dissolution rate and is more stable at high pressures. [39] Ethenzamide-2,4 dihydroxybenzoic acids 1:1 Neat grinding Has good stability, non-hygroscopic, higher solubility. [40] Ethenzamide-saccharine 1:1.1 Neat grinding Significantly increases solubility and permeability. [41] Ethenzamide-sinapic acid 1:1 Solvent evaporation Has good thermal stability. [42] Diflunisal-nicotinamide 2:1 Solution co-crystallization Thermal stability, improvements to solubility and dissolution. [43] Diflunisal-isonicotinamide 2:1 Solution co-crystallization Diflunisal-pyrazine 2:1 Ball milling [44] Diflunisal-theophylline 1:1 Solvent-drop grinding Increased solubility up to 2.3-fold, indicated for asthma therapy. [45] [47] Indomethacin-2-hidroksi-4metilpiridin 1:1 Slow solvent evaporation Increased permeability up to 2-fold. [15] Heteroacil acetic acid Diclofenac-L-proline 1:1 Liquid assisted grinding and neat grinding Increased stability and solubility up to 7.69-fold. [48, 49] diclofenac-theophylline 3:1 Solvent-drop grinding Increased dissolution rate up to 1.3-fold, indicated for asthma therapy. [45] Diclofenac-ethyl diclofenac 1:1 Increased chemical stability and inflamatory activity. [50] Sodium-diclofenac-Lproline-tetrahydrate 1:1:1:4 Increased stability and solubility up to 3.1-fold. [27] Sodium-diclofenac-Lproline-monohydrate 1:1:1:1 Increased dissolution rate and solubility up to 4.8-fold. Potassium-diclofenac-Lproline-hydrate 1:1:1:4 Increased solubility and intrinsic dissolution rate salt potassium diclofenac up to 3.55-fold. [51] Nano-diclofenac-L-proline 1:1 Top-down with the ng and bottom-up with microwaving assisted rapid evaporation Increased solubility up to 2.64-fold and increased diffusion up to 2.06 fold that of the dfa-pro co-crystal. [52] Aceclofenac-nicotinamide 1:1 Neat grinding Reduced size, increased bioavailability up to 1.77-fold. [31, 53] Aceclofenac-gallic acid 1:1 Solvent evaporation Increased bioavailability of co-crystal up to 1.73-fold. [54] Aceclofenac-lysine 1:1 Solvent drop grinding Higher melting point than parent drug, 201 • C [55] Arylpropionic acid Ketoprofen-nicotinamide 2:1 Solvent evaporation Increased solubility up to 1.3-fold. [56] Naproxen-nicotinamide 2:1 Supercritical antisolvent Size distribution ranging between 20 µm and 1 mm and flow ratio increased to 36 times. [57] Naproxen-urea 1:3 Solvent evaporation Increased dissolution rate up to 4.37-fold, has good powder flow properties and compressibility. [58] Naproxen-thiourea 1:2 Solvent evaporation Increased dissolution rate up to 4.5-fold. NSAID co-crystals exhibit various supramolecular synthon motifs ( Figure 2 ) [84] . For example, diflunisal (DIF) co-crystal has three hydrogen bonding patterns based on the Cambridge Structural Database (CSD), namely (1) a ring hetero-synthon with aromatic N, (2) a COOH-COOH homo-synthon, and (3) four ring homo-synthons between OH and C=O from acids. Interestingly, DIF and other salicylic acid (SA) derivatives have an ortho-hydroxy group with steric resonance and hydrogen bonding called the "ortho effect". Only eight DIF co-crystals have been created with the following features: (a) a COOH homo-synthon between acid co-formers and o-hydroxy carboxylic acids when one of them has an electron-attracting group, (b) COOH homo-synthons between acid co-formers and o-hydroxy carboxylic acids when the o-hydroxy acid has a competing group, such as OH or NH 2 , which can form hydrogen bonds with the co-former, and (c) a combination of motifs (a) and (b) with a COOH interaction distance of more than 3 Å; the type (c) motif is rare. Based on their feature exploration, the ability of donors is in the following order: COOH > NH > OH with the most interactions found in DIF with COOH [38] . Indomethacin (INC) also forms a co-crystal with the saccharine (SAC) co-former by forming N-H·· O hydrogen bonds between acid dimers in INC and imide dimers in SAC [48] . Naproxen (NPX)-picolinamide (PA) co-crystal was formed from carboxylic acidcarbamide dimers between NPX and PA, as shown in Figure 3 . PA offers an ortho effect with a pKa value of 1.17. For that PA, carboxamide interacts with NPX via acid-amide synthesis; however, SCXRD results do not match the asymmetric unit. The position of the proton determined by H nuclear magnetic resonance (NMR) shows that the H atom involved in hydrogen bonding at the synthon gives a different peak. The H atom involved has a delay of 20s compared to the other H atom, which was determined by comparing the NMR data with the predicted shift values. The RMSD obtained was 0.09 Å by optimizing the atomic position at 0.2 Å SCXRD so that the hydrogen bonds are more symmetrical and the difference between the two hydrogen bonds O-H · O was reduced. [85] . Adapted from ref. [85] . Copyright 2017 NCBI. In Figure 3c , two bonding models are proposed as alternatives to improve the SCXRD results. In model 1, only the H atoms in the OH bond are removed, whereas, in model 2, the amide protons in PA also move along the N-H hydrogen bond. H atom position was validated by density functional theory (DFT) geometry optimization based on a significant decline in potential energy, which indicates that the naproxen-picolinamide (NPX-PA) is a co-crystal and not a salt [85] . Lornoxicam (LNX) is included in the oxicam class of NSAID drugs. LNX has more selective anti-inflammatory activity without affecting the digestive tract. LNX can be used for osteoarthritis therapy [86] and postoperative pain relief of the head and neck, and also reduced postoperative pain in the third molars in the first phase of perioperative pain management [87] . LNX has better tolerance and safety than tramadol and is comparable to tramadol in postoperative analgesic therapy [88] . Oral administration of LNX takes 2.5 h to produce a maximum concentration (Cmax), limiting LNX therapy for providing the optimal effect as an analgesic. The co-crystallization technique has succeeded in changing the physicochemical properties of LNX using the liquid-assisted grinding (LAG) method. From this method, co-crystals of LNX salt were produced with a benzoic acid co-former and 2,4 dihydroxy benzoic acid. In contrast, co-crystals were formed from catechol co-formers, resorcinol, hydroquinone, and SAC at a stoichiometric ratio of 1:1 [77] . The crystal system is orthorhombic with space group P 212121 with one LNX molecule in asymmetrical units based on the synthon co-crystal approach. The decrease in the melting point of the parent drug indicates the formation of a multi-component system. Strong hydrogen bonds link the LNX N + -H···O=C and NH···OC causes a decrease in the conformational change in the amide bond to the ring. As a result, a zigzag band formed along the C axis in the molecule through the interaction of N + -H···O=S, C-H···O and C-H···Cl, then the CH···S interaction forms a 2D sheet and creates a 3D layered structure through CH···O hydrogen bonding interactions. Based on in vitro dissolution testing at the optimum temperature of LNX, the salt co-crystal has excellent solubility compared to the co-crystal and LNX, i.e., up to 1.52-fold higher than that of LNX [78] . LNX co-crystals were also formed by the neat grinding method on the generally recognized as safe (GRAS) co-former and drugs such as malonic acid (MAL), succinic acid, tartaric acid, anthranilic acid (phenamate), cinnamic acid, p-aminobenzoic acid, ferulic acid, urea (URE), sodium saccharin (SS), citric acid, and OXA. LNX-SS increases the solubility of the parent drug significantly. It is related to the low lattice energy value, thereby reducing the interference from the solvent. Co-crystallization decreases the partition coefficient and shows the transformation from a hydrophobic to a hydrophilic compound. The LNX-SS co-crystal has good stability at extreme temperatures based on stability testing for 30 days under 40 • C and 75% humidity [76] . Co-crystal diclofenac (DFA) with pyridine-based co-formers and acid pyridine synthon has been reported in nine new solid forms using the evaporation co-crystallization technique. DFA formed co-crystals with the 1,2-bis (4-pyridyl) ethane (BPE), 1,3-di (4-pyridyl) propane, and 4,4-bipyridine at a stoichiometric ratio of 2:1. The three co-formers have pyridine groups on both sides. The packing of DFA co-crystal is highlighted in Figure 4a ,b. Simultaneously, DFA formed salt co-crystal with two amino groups as co-formers, namely 2-aminopyridine (2-apy) and 3-aminopyridine. In the salt co-crystal DFA-2-apy, there is a weak bonding of C-H. There is a tetramer aggregate packaging in the salt co-crystal that connects the synthon, as shown in Figure 4c [25] . The formation of NSAID co-crystals with an amino acid as the co-former has been widely published, called zwitterionic co-crystals. For example, indomethacin-proline (INC-PRO) 2H 2 O, a co-crystal hydrate, exhibited the supramolecular synthons NH 3+ ···O − and OH···O. The INC-PRO co-crystal consists of an internal hydrophilic group and a hydrophobic group on the surface, increasing the solubility and permeability INC [23] . Besides INC, a diclofenac-proline (DFA-PRO) was obtained as a stabilized co-crystal and linked with intramolecular hydrogen bonds N-H···O and C-H···C. The difference in the C-O bond length of the DFA and PRO carboxylate groups > 0.07 indicates that DFA-PRO is a zwitterion co-crystal [48, 49] . Furthermore, Nugrahani et al. developed the DFA-PRO co-crystal by forming a salt hydrate co-crystal to increase the multi-component solubility and stability over DFA-PRO co-crystal and the alkaline salt diclofenac. Besides sodium diclofenac proline tetrahydrate (SDPT), a co-crystal of sodium diclofenac-proline monohydrate (SDPM) was confirmed by a structural determination with SCXRD at −180 • C. The new phase consisted of diclofenac sodium-proline-water (1:1:1:1). A stability study of the two co-crystals was carried out under extreme drying and humidity and evaluated by differential thermal analysis (DTA)/TG/DSC). Based on the diffractogram results, SDPT was stable at high humidity, i.e., 94 ± 2%RH/25 ± 2 • C, for up to 15 days did not separate into the starting materials [48] . Meanwhile, SDPM quickly changed to SDPT under these environmental conditions. Moreover, SDPM could be restored to SDPT under dry conditions, so SDPT is a more stable co-crystal than SDPM. These results show that water plays a critical role in forming co-crystals in SDPT to mediate the interactions between the components in the co-crystals of the tetrahydrate salt formed. The high solubility of SDPM occurs because there is a region consisting of Na + and water molecules with high affinity, such that it quickly dissolves and breaks down. The dissolution test was carried out at pH 1.2 and 6.8, based on previous studies; the sodium diclofenac (SD) profile increased at moderate to alkaline pH. Based on this test, the SDPM and SDPT dissolution results were faster than SD, with a superior increase in the dissolution of SDPM. This result showed that the monohydrate crystal lattice was smaller and had a looser and broader space than the tetrahydrate form, so it dissolved faster than SDPT at pH 6.8 [27] . Furthermore, the DFA-PRO co-crystal was developed from diclofenac potassium, resulting in a salt co-crystal hydrate consisting of potassium, DFA, L-proline, and water (1:1:1:4) [51] . The prediction of bond formation between co-crystal NSAID and co-formers using a computational method to calculate DFT with B3LYP/aug-cc-pVDZ 1 depends on the constituent molecules to form a supramolecular synthon [44] . Co-crystal screening with various co-formers and methods has been established, one of which is by calculating the van Krevelen and Hoftyzer solubility parameters, involving the solubility of two compounds to find the ∆δ − factor with the equation: The partial solubility parameters are dispersion (δ d ), polarity (δ p ), and hydrogen bonding (δ h ). Good miscibility can be achieved if ∆δ ≤ 5 MPa 0.5 . The difference in total solubility between the drug and the ∆δt carrier is a means of predicting solubility: As noted, t1 and t2 are carriers and drugs; materials are miscible with ∆δ ≤ 7 MPa 0.5 , while systems with ∆δ ≥ 7 MPa 0.5 are immiscible. This solubility parameters method was used for developing aceclofenac (ACF) co-crystals with several co-formers, namely gallic acid (GLA), CA, MLA, NIC, D-tartaric acid (TCA), URE, and vanillic acid, which showed that all co-formers could form co-crystals, except TCA [31] . ACF 2-[2-[2-[(2,6dichloro-phenyl)amino]phenyl]-acetyl] oxy-acetic acid (C 16 H 13 C l2 NO 4 ) is an analog of DFA in the form of glycolic acid ester. It is used as a first-line drug for the treatment of rheumatoid arthritis and osteoarthritis [89] . The gastrointestinal side effects of ACF are relatively lower than other non-selective NSAID and are comparable to CEL [31] . ACF does not interact directly with COX enzymes, but ACF in the form of a prodrug will produce DFA active metabolites, which will inhibit COX enzyme activity by forming hydrogen bonds on the TYR355 and SER530 residues of COX enzymes [90] . The characterization of the ACF-NIC and ACF-GLA co-crystals obtained by the solvent evaporation technique at 1:1 stoichiometry was based on the SEM analysis. The ACF-GLA and ACF-NIC co-crystals had different crystal shapes than pure ACF, large and regular, while the co-crystals had a more irregular shape [31] . Based on the FTIR structure, five hydrogen bonding motifs arranged the ACF-NIC co-crystal, which is (1) the formation between the amide group from NIC and the acid group from ACF, (2) hetero-synthon chloride-amide formation, (3) ACF linked to NIC by an acid-pyridine hetero-synthon (synthon I) or acid-amide hetero-synthon (synthon V), (4) amide-amide dimers by NIC and ACF groups bound to an acid-amide hetero-synthon, and (5) ACF formed by synthon I and an amide-chloride hetero-synthon [53] . In developing co-crystals, the method is crucial to produce the expected physicochemical properties, as demonstrated by the indomethacin-proline (INC-PRO) system, produced by LAG, and solvent evaporation. The LAG method yielded polar molecules, thereby improving the solubility and intrinsic dissolution rate (IDR) under each tested pH condition; high solubility indicates increased bioavailability and may improve the pharmacokinetic profile of INC [23] . Jafari et al. categorized the co-crystal production route in two ways, solution and solid-state based. Evaporative co-crystallization, cooling crystallization, reaction co-crystallization, isothermal slurry conversion, and supercritical antisolvent (SAS) are solution-based production methods. Conversely, the solid-state process is a technique to combine the solid materials directly, and pressure is applied manually (mortar and pestle) or mechanically (automatic ball mill). The most common solid-state methods are neat (dry) grinding (NG), LAG, and hotmelt extrusion. The direct solid-state process significantly reduced solvent usage, so it is preferable in the context of the evergreen method [91] . However, the different techniques may not produce the same form and physicochemical properties, evidenced by several publications. For example, Evora et al. reported a co-crystal with three methods: annealing a mortar ground mixture at 80 • C, annealing at room temperature after neat ball mill grinding, and ethanol-assisted (10 µL) ball milling. The neat mill grinding methods produced only a few crystals. Meanwhile, the annealing mixture at 80 • C formed the new solid phases with a melting point at around 113-146 • C. In this study, eight co-crystals were created with the ball milling method. In contrast, the diflunisal (DIF) co-crystal with pyrazine, which has a crystalline form like a diamond, was formed by a crystallization method using a solution. DIF can be crystallized with a greener solvent with OXA as the co-former [87] . The solution method was then compared to LAG with ethanol and NG. Observation with FTIR at a specific time showed significant changes in the spectra (Figures 5 and 6) [49] . With a longer grinding time, the specific peak transformation of the co-crystal was more apparent as the peak of the parent compound decreased and disappeared. The regions at wavelengths 1500-2500 cm −1 and 2500-4000 cm −1 show the specific evolutionary areas on the formation of the co-crystal, which indicates a gradual shift of the carbonyl groups to a lower wavenumber, i.e., from 1693 cm −1 to 1685 cm −1 after 30 m of the grinding process. This phenomenon is related to the decreased vibrational energy of the carbonyl, which is caused by the formation of heterocyclic N hydrogen bonds with the PRO group. Besides the carbonyl, the OH peak also experienced a shift from 3324 cm −1 to 3270 cm −1 accompanied by a new band at 3170 cm −1 , estimated from the OH carbon oxide group DFA which forms hydrogen bonds with the PRO carbonyl group. Therefore, this shift was accompanied by a change in the PRO carbonyl group from 1652 and 1617 cm −1 to 1623 and 1616 cm −1 . The new peak appeared at 1968 cm −1 due to the presence of hydrogen bonds O···H-N, which occurred after 2 min of grinding and formed a sharp spectrum after 60 min of grinding. The specific peaks of the co-crystal appeared regularly at 3270 and 3170 cm −1 . The DFA-PRO co-crystal dynamics using the NG method showed the same FTIR pattern as the LAG method; however, the initial co-crystal formation took longer in co-crystallization using the NG method, i.e., within 10 m, while with the LAG method, the initial co-crystal was formed within 2 min. This result is consistent with previous research showing that the addition of solvents will increase molecular diffusion, thereby increasing the interaction between the API and co-formers [49] . Sevukarajan et al. successfully synthesized an ACF-NIC (1:1) co-crystal with the NG method which showed a better solubility because it produced smaller crystals than the solvent evaporation (SE) method [43] . Besides offering a different solubility, different co-crystallization methods also yield various crystalline forms. Berry et al. obtained two co-crystal phases of ibuprofen-nicotinamide (IBU-NIC) by melting and slow evaporation method, namely the rectus-sinister ibuprofen-nicotinamide (RS-IBU-NIC) and sinister ibuprofen-nicotinamide (S-IBU-NIC) [92] . Guerain et al. also studied the formation of IBU-NIC co-crystal with different co-crystallization methods: (1) milling, (2a) crystallization by melting the mixture at 100 • C and then cooling to room temperature, (2b) crystallization by melting the mixture at the glass transition temperature, and (3) the slow evaporation of the solvent. After being characterized by PXRD, each co-crystallization method produced different S-IBU-NIC co-crystals [93] . Methods (1), (2b) and (3) resulted in a similar S-IBU-NIC found by Berry et al. In contrast, co-crystallization with a process (2a) resulted in a new S-IBU-NIC co-crystal. The stability of the new S-IBU-NIC co-crystal showed transformation to the previous S-IBU-NIC co-crystal at temperature 65 • C. In this study, the crystallization method by heating close to the melting temperature will produce polymorphisms [93] . Mefenamic acid (MFA) can arrange co-crystals with several co-formers, such as NIC, URE, pyridoxine, etc. Of all the co-formers reported, NIC was selected as the co-former used in a study on co-crystal formation using the melt crystallization technique because contaminants can form when MFA is co-crystallized by a grinding process. This technique was carried out by melting MFA with NIC (1:2) in a porcelain cup over a paraffin oil bath with the temperature maintained at 200 • C, then incubating in a container containing water over a water bath with the temperature maintained at 90 • C, then drying at room temperature overnight. Based on the characterization results with PXRD, FTIR, DSCthermogravimetric analysis (TGA), and thin-layer chromatography (TLC), the heating process did not cause co-crystal decomposition. Moreover, from the solubility test results, the solubility of MFA-NIC was higher than that of the parent compound [64] . The supercritical anti-solvent (SAS) method has been used in the formation of NSAID co-crystals. The principle of this method is to dissolve the API and co-former until saturated with the appropriate solvent. The addition of CO 2 reduces the solubility of the API and co-former to induce precipitation to produce NSAID co-crystals [91] . SAS methods can also compose naproxen (NPX)-NIC co-crystals at an equimolar ratio of 2:1 with a solvent mixture (acetone and CO 2 ) subjected to high pressure (10 MPa Figure 7 . PXRD and FTIR analysis concluded that the SAS technique produced NPX-NIC co-crystals with the same characteristics as conventional co-crystallization techniques [57] . Wichianphong et al. formed MFA-NIC co-crystals using the gas anti-solvent (GAS) method under conditions optimized with the Behnken experimental design. Parameters that can affect the process were investigated, such as temperature (T), the molar ratio of the co-former to the drug (C), and the percentage of drug saturation (S) in the solution against t 63.2 (the time required to dissolve 63.2% of the drug). The resulting co-crystals were then compared also with the product from the conventional methods and physical mixtures. Based on the difference in melting point, PXRD pattern, and FTIR spectrum, the formation of a new phase was confirmed, namely the MFA-NIC co-crystal. The fastest dissolution time (5.07 min) at 450 • C, a co-former-to-drug ratio of 5:1, and 70% drug saturation. The t 63.2 obtained from the experimental results and the calculation of the equation was suitable and met the requirements of (R 2 ≥ 80%) with R 2 amounting to 96.25% and 89.51%; hence this model provides a good correlation. ANOVA showed that S and C significantly affected the t 63.2 (p < 0.05 and value of F = 14.27) where, at low T, t 63.2 could reach the minimum value if S and C were high, which can increase the S value. A high S value can prevent the formation of MFA-NIC co-crystals and result in a faster dissolution time [65] . In addition to DSC, PXRD, and FTIR, the co-crystal MFA-NIC co-crystallization results by the GAS method were also characterized by SEM. Observations showed that the form of MFA-NIC was different from the parent compound and the co-crystal produced by the conventional method ( Figure 8 ). From the dissolution test, the co-crystals from both methods provided an increase in the dissolution rate ( Figure 9 ) [65] . The advantage of this method is a uniformly shaped co-crystal with high purity due to the filtration process with a particular size and a washing process with CO 2 to remove the remaining solvent in the MEF-NIC co-crystal [65] . In addition, this SAS technique depends on the composition of the fluid used and can cause heterogeneity [57] . As the development of technology in pharmaceutical engineering increases, the cocrystallization technique continues to be developed to improve solubility and permeability, one of the premises of applying nanotechnology in co-crystallization. Nanotechnology is the most established technology for increasing the therapeutic index and overcoming the challenges of formulating compounds with poor solubility. From several studies, the nano co-crystal approach has been shown to increase the solubility and dissolution of drugs with poor solubility. This result is related to the size of the co-crystal being reduced to nano-size (less than 100 nm), increasing the surface area and the dissolution rate. Increased dissolution will be very useful to enhance bioavailability. Moreover, nano co-crystal engineering can reduce toxic solvents and surfactants and develop a formulation for various routes of administration where size is a critical factor (injection, ophthalmics, and topicals) [95] . The nanocrystal mineral has limitations, including that it can only be used for class II BCS drugs, requiring high-cost instruments, with formation and stability depending on the drug molecule. Only certain compound groups meet the requirements. A large nano-co-crystal surface with high free energy or differences in surface charge can cause aggregation. However, an increase in solubility that exceeds the saturation point will cause recrystallization into larger particles, called Ostwald refining. Selecting a stabilizer in the formulation of a nano-co-crystal preparation to protect the particle surface in order to reduce the free energy of the system and the interface voltage of the nano-co-crystal overcame this occurrence [96] . The nano-diclofenac-proline co-crystal (NDPC) is formed by combining two techniques, i.e., top-down with the NG method and bottom-up with microwave-assisted rapid evaporation. The production of NDPC with NG produced pure nano-co-crystals with a particle size of around 857.9 nm with a polydispersity index (PI) of 0.353 after 6 h. However, there are limitations to this method related to co-crystal instability. The best sized NDPC at 598.2 nm and PI 0.278 was successfully obtained by the fast evaporation method with ethanol as the solvent and 776 W microwave energy in 8-12 min. Microwave energy provides molecular rotation to form intermolecular bonding interactions in the co-crystal. In stabilizing NDPC, sodium lauryl sulfate (SLS) was added as a stabilizer, giving a zeta potential value of −660 mV, indicating the particles are stable without forming agglomeration. In addition, SLS provides a negative charge on the zeta potential to disperse the nano co-crystal solution [52] . A scale-up process was carried out to produce pure crystals on the scale of 10 g using the refrigerant crystallization method without seeding (adding crystal seeds), equipped with a temperature sensor. This method used a heating and reflux technique to dissolve the co-crystal in ethyl acetate, slowly lowering the temperature with an optimized stirring cycle. Based on PXRD and DSC analysis, the scaled-up indomethacin-saccharine (INC-SAC) co-crystal product has the same purity with the yields of a small scale as the solvent evaporation method using ethyl acetate [47] . Apart from these methods, a combination method (SAS and cooling co-crystallization) could scale up the INC-SAC co-crystal. Cooling time is a factor for optimal production conditions, as cooling will accelerate the precipitation of the co-crystal. The post-nucleation cooling time resulted in a greater amount of INC-SAC co-crystal with smaller particle sizes [97] . Based on studies conducted by Skorupska et al. to increase solubility and prevent degradation, modified drug delivery was studied by inserting the naproxen-picolinamide (NPX-PA) co-crystal into mesoporous silica particles (MSP) using the thermal solventfree method, by heating a mixture of co-crystal and MSP at 100 • C for 2h. The complex NPX-PA co-crystal in MSP as shown in Figure 10 was prepared to protect the API from environmental effects, carry the drug across the cell membrane, accelerate the drug's action, increase treatment efficiency, and deliver the drug to specific target organs. Two MSPs with different pore sizes were used; NPX-PA was successfully inserted into SBA-15 (100 Å), while MCM-41 (37 Å) acted as a separating medium. NPX stuck to the outer wall of MSP since the pore size of MCM-41 is smaller than NPX, so SBA-15 is more suitable for the NPX-PA co-crystal [36] . Sohrab et al. conducted a study to observe the effect of a water-soluble polymer on aceclofenac-nicotinamide (ACF-NIC) co-crystal 1:1 obtained from the solvent evaporation and NG methods. PVP K30, hydroxypropyl methylcellulose (HPMC), sodium starch glycolate, and carboxymethylcellulose sodium (CMC-S) were mixed with the co-crystals, and the mixture was then tableted by the wet granulation method. The addition of watersoluble polymers lowered the melting point of the co-crystals, decreased the lattice energy, and thus increased the dissolution rate of ACF. Based on dissolution testing with the USP type I method in phosphate buffer pH 7.5 medium, the ACF-NIC co-crystal without the addition of 3% water-soluble polymer showed 78.9% and 78.5% drug release, while the co-crystal formulations MUNG01 (3% PVPK-30), MUNG02 (3% HPMCE5), MUSE03 (3% SSG) and MUSE04 (3% CMC-S) showed maximum release of 99.1%, 97.51%, 99%, and 98.25% [98] . Next, ACF was also developed into a topical formulation, which was expected to have a high penetration rate through each layer of the skin to effectively and safely treat pain locally. Sharma et al. developed a nanoliposome formulation of ACF with amino acid lysine (LYS) to increase the co-crystal penetration to the skin. ACF-LYS was encapsulated using liposomes formed from lipids and cholesterol in a 70:30 ratio, then dissolved in chloroform and evaporated. ACF-LYS was suspended in the liposome and then inserted into a vehicle from carbopol 940 hydrogel to increase viscosity and form a translucent gel. The co-crystal-loaded liposome gel (COC-LG) formulation, with an average size of 120.4 ± 1.03 nm, showed greater drug penetration into the deeper skin layers. The co-crystal had a spherical shape based on imaging studies with confocal laser scanning microscopy and transmission electron microscopy. The COC-LG formulation had a lower viscosity than the gel on the market, which indicated good contact time and dispersion, resulting in maximum therapeutic effect. Based on ex vivo testing using the back skin of Wistar rats, the formula penetrated better and remained in the skin 2.31 times longer than the marketed gel. The penetration increase was associated with the interaction of the biocompatible component (phospholipon) with the skin. The COC-LG formulation increased the drug concentration in a short time, did not interfere with skin integrity, did not cause inflammation of the dermal tissue (as did other gels), and was more effective as an analgesic and anti-inflammatory for the treatment of arthritis and osteoarthritis [55] . Co-crystallization can also consist of APIs, metals and organic materials to accelerate NSAID action and prolong the half-life, thereby increasing the drug's duration of action. Hartlieb et al. formed an IBU-metal-organic co-crystal called a metal-organic framework (MOF). MOF is a γ-cyclodextrin (γ-CD) metal used in the form of cation metals such as potassium (K + ) or sodium (Na + ). CD-MOF was used to create an organic porous framework filled with IBU. The co-crystals produced by the diffusion of ethanol vapor in a CD solution and ibuprofen potassium salt formed a cyclodextrin metal-organic framework (CD-MOF-1) co-crystal with an IBU loading rate of 23% in the CD-MOF-1 co-crystal. One problem with IBU is the active S-enantiomer and less active R-enantiomer. However, IBU was absorbed by CD-MOF-1 as the less active enantiomer, but MOF succeeded in separating the enantiomer from IBU. Co-crystal CD-MOF-1 had better stability in the atmosphere than IBU salt and was not hygroscopic, as seen from the PXRD pattern, where it did not show co-crystal degradation in ambient humidity. Based on bioavailability testing with female rats, the pharmacokinetic data showed that the CD-MOF-1 co-crystal had a Cmax, area under the curve (AUC), and a half-life that was two times higher than that of the IBU salt. This result indicated that IBU absorption was fast with an onset of 10-20 min [99] . Co-crystallization of two APIs was also arranged to improve the physicochemical properties of the drug combination. This strategy benefits NSAID drugs, where their therapeutic use is often connected to treat mild to moderate pain or for continued treatment of gout if single therapy cannot be used. Pathak et al. performed co-crystallization techniques on PCM, INC, and MFA drugs with various methods such as solvent evaporation, grinding, the addition of antisolvents, and ultrasound-assisted techniques. Co-crystals were formed from solvent evaporation, which worked best compared to other techniques. The PCM-INC and PCM-MFA co-crystals created hetero-synthon supermolecular synthons with strong hydrogen bonds between COOH-N and COOH-O [83] . The formation of NSAID drug-drug co-crystals can improve drug stability, i.e., the piroxicam (PXC) co-crystal with an aniline-nicotinic acid derivative NSAID, namely clonixin (CNX). Based on stability testing, there was no change in color and transformation into a hydrate at a storage temperature of 25 • C and humidity up to 95% for 4 weeks. Co-crystallization of the PXC-CNX aimed to inhibit the phase transition from the anhydrous to hydrated form, changing the physicochemical and pharmacokinetic properties. CNX was chosen because it has a synergistic therapeutic effect with PXC and CNX structure and has many carboxyl groups to be a co-former. Moreover, it can form both neutral and zwitterionic molecules. In the formation of the PXC-CNX co-crystal, three parameters need to be considered, i.e., the hydrogen bond donor and acceptor's ability and the polarity of the solvent used. The moderate polar solvents can be used to produce the PXC-CNX co-crystal. Moreover, thermal analysis with DSC showed that the solvent molecule played an essential role in stabilizing the crystal structure. The calculation of the interaction energy with DFT showed that homo-molecular interaction has superior energy than hetero-molecules. Homo-molecular bonds act as a driving force for the decomposition of PXC-CNX compared to a solvent-free co-crystal. Based on screening with the slurry method, there was only a difference in the ethyl acetate (EA) solvent. The PXC-CNX-EA had a different C-H···C distance EA molecule, longer than other solvent molecules. In addition, EA is a non-toxic compound, so that it is safe for food and drug formulations. The crystal structure of PXC-CNX was obtained after the solvent was evaporated and the two molecules were linked via O-H···O hydrogen bonds between the COOH group on CNX and the OH group on the deprotonated PXC. The PXC zwitterion formed hydrogen bonds with the CNX zwitterion via N-H···O, comprising a protonated pyridine donor and a carbonyl amide acceptor [71] . Combined NSAID have better anti-inflammatory activity than single drugs, such as DFA-ethyl diclofenac (ED) co-crystalline, which is more effective than diclofenac acid because only a small amount of ED is bound to plasma. However, the esterification process between the ester and DFA can produce unwanted products. Therefore, a co-crystal between ED and DFA was made, and crystallographic studies were carried out. From the results of thermal analysis with DSC, the melting point of the co-crystal was greater than that of ED (67.7 • C) and lower than that of DFA (173.1 • C), i.e., around 103-104.3 • C. The results showed that the FTIR spectrum did not change. Thin layer chromatography analysis showed the same pattern for the two accelerated stability tests, heated with microwave energy and stored in high humidity for seven days. Thus, the ED and DFA co-crystal were chemically stable. Moreover, anti-inflammatory activity was studied using five groups of mice, which showed that the co-crystals between ED and DFA increased anti-inflammatory activity more than the single components [50] . SCXRD, PXRD, and FTIR characterized crystal meloxicam (MLX) and aspirin (ASP), showing different spectra from the starting material. Based on the rules of pKa, the meloxicam-aspirin (MLX-ASP) co-crystal has not been defined, with a pKa value of 0.68, and a neutral phase with a CNC angle of 110.2 • . In a comparable solubility study in the testing medium of phosphate buffer solution pH 7.4 at 37 • C, the solubility of MLX in MLX-ASP co-crystals increased to 0.22 mg/mL from 0.001 mg/mL. Pharmacokinetic studies using male Sprague-Dawley rats at an oral dose of 1 mg/kg showed an increase in absorption of the MLX-ASP co-crystal with a bioavailability of 69%. At the same time, that of MLX was only 16%. The MLX-ASP co-crystal could also achieve tmax four times faster than the single MLX, so it reached the concentration of 0.51 g/mL MLX in 10 min. This rapid onset is an advantage of the MLX-ASP co-crystal, which is indicated as a mild to moderate acute pain reliever. In this study, ASP in the MLX-ASP co-crystal was only 7.7 mg and MLX 10 mg/kg, so it did not cause significant side effects [74] . The multi-APIs co-crystal was also constructed between celecoxib (CEL) and tramadol HCl (TML···HCl), named CTC. Initially, the co-crystal was formed by grinding method using isopropyl alcohol. This new phase is an ionic co-crystal, where the chloride ion interacts with an adjacent drug molecule by involving -N + H···Cl − and -OH···Cl hydrogen bonds. The maximum concentration of CEL in CTC was higher than the maximum concentration of a single CEL, thus extending drug action by slowing drug release. Additionally, this combination increased efficacy, preventing pain by four mechanisms of action and lower therapeutic doses, as the CTC dose was 100 mg (44 mg TML · HCl and 56 mg CEL). In contrast, the dose of CEL and TML · HCl is around 200-400 mg/day [79] . Because of its very high efficacy, the co-crystal (TML · HCl-CEL) was tested in a phase I clinical trial in 2017 by Esteve (E-58425) and Mundiphara Research (MR308). In stage I, a single pharmacokinetic (PK) dose of CTC and its reference products were compared singly and in combination. Samples were randomly assigned with an initial dose of 200 mg of CTC, equivalent to 88 mg of tramadol and 112 mg of CEL. The results of the AUC value were comparable to the Cmax, but lower than that of tramadol. However, the t-max was longer than that of tramadol. This result is consistent with previous research where C-max reduction was proportional to the slow dissolution rate [100] . Next, they continued the phase II clinical trial in patients. There were six doses of CTC tested (CTC 50, 100, 150, 200 mg; tramadol 100 mg; and placebo). The initial efficacy of this test was based on differences in pain intensity across the population. Based on the results of phase II, the potential for CTC was more significant than the risk. Furthermore, CTCs at 100, 150, and 200 mg doses were more effective than tramadol 100 mg and placebo for treating moderate to severe acute pain [80] . CTC development is currently awaiting the results of a phase III trial [101] [102] [103] , and this is the first co-crystal NSAID to enter clinical trials, while the MLX-ASP co-crystal is still in the in vivo testing phase [74] . For gout therapy, NSAID are the first-line drugs to reduce pain, which need to minimize uric acid levels in the body by increasing excretion or inhibiting the formation of uric acid using uricosurics and xanthine oxidase (XO) inhibitors [89, 90] . To achieve optimal therapy, Modani et al. utilized co-crystallization in combining PXC with febuxostat (FBX). FBX is a novel non-purine xanthine oxidase (XO) used to treat hyperuricemia in gouty patients. Based on several studies, FBX is effective at inhibiting lung inflammation in animals caused by oxidative stress [104] [105] [106] , has been shown to accelerate restoration of the pulmonary endothelial barrier [107] , and is potent in treating mild to moderate COVID-19 infection by overcoming the early-phase pneumonia caused by the coronavirus [108] . A single crystal of the piroxicam-febuxostat (PXC-FBX) co-crystal with a stoichiometric ratio of 1:1 resulted from the crystallization technique using acetonitrile solvent. The formation of co-crystals was confirmed by a decrease in the melting point, no stretching of the OH from carboxylic acids indicating no proton transfer, the specific shape, and differences in the diffractogram. A supramolecule synthon connects the carboxylate and azole groups with hydrogen bonds NH···O and 2D packaging, stabilized by the interaction of NH···O, CH···N, OH···O, and CH···O hydrogen bonds. It becomes interesting to learn when both APIs come from the same BCS class. Solubility testing at pH 1.2-7.4 resulted in 2.5 times higher solubility for FBX at pH 1.2 and continued to increase as pH increased; however, at pH 4.5, the solubility decreased. At pH 6.8 and 7.4, the solubility of PXC significantly increased without affecting FBX, so that at pH 1.2 PXC was a co-former for FBX but at pH 6.8 and 7.4, FBX was a co-former for PXC. In vitro, the FBX-PXC co-crystal improved the dissolution time by up to 2.8-fold compared to pure PXC without affecting the dissolution of FBX. In addition, the FBX-PXC co-crystal had a stable crystal structure and altered mechanical properties such as an increase in the flow rate and the formation of capping and laminating at the time of compression, showing good compressibility [72] . NSAID, especially the anthranilic acid group, also can construct multi-APIs co-crystal with antibacterial drugs. This combination aimed to provide a unique combination drug or pain relief therapy and prevent postoperative infections. Hence, it minimizes or even replaces the use of opioids, reduces adverse effects, and improves the physicochemical properties of each API [30] . For example, niflumic acid (NFA)-caprolactam (CPR) (1:1) is an example of an anthranilic acid class of NSAID that selectively inhibits COX-2 and is widely used in patients with rheumatoid arthritis. The co-formers used to form NFA co-crystals are CPR and 2-hydroxy pyridine (2HP). An analysis of the crystal structure showed the type of crystals that form between the salt and the co-crystal together by bonding through NH···O and OH···O hydrogen bonds, which form dimers, and CH···F and CH···O. The NFA-CPR co-crystal has a low melting point of 83 • C. In contrast, the NFA-2HP co-crystal has a melting point of 135 • C [69] . Bhattacharya et al. performed a screening of the hydrate salts and multi-API cocrystals of anthranilic acid with an antibacterial. There are two co-crystals from APIs, namely flufenamic acid (FFA) with sulfamethazine (SFZ) and niflumic acid (NFA) and SFZ. In the crystal structure, hydrogen bonds are formed between the carboxylic acid groups of NFA/FFA with the sulfonamide (NH) groups and N atoms from the pyrimidine ring on the SFZ to form synthon III. Then, each dimer unit is connected by four pairs of identical NH···O hydrogen bonds to form synthon IV. This structure is stabilized by CH···interactions and π···π interactions [30] . Machado et al. conducted a study on an NSAID co-crystal formed from propionic acid with levetiracetam (LEV), which was expected to treat epilepsy accompanied by inflammation. LEV is an etiracetam enantiomer and acts as an oral anti-epilepsy drug classified as BCS class I (good stability and permeability), so it is expected to improve the physicochemical properties of the NSAID. The eutectic mixture and co-crystal dissolution test with aryl propionic acid (IBU, naproxen (NPX), Flurbiprofen), and LEV showed an increased dissolution rate compared to the pure NSAID. LEV + (S)-IBU crystals were stable in accelerated stability tests over six months. However, the eutectic mixture of propionic acid and LEV melted at temperatures below that of the pure NSAID. This suggests that the crystalline form of propionic acid and LEV can also increase propionic acid derived NSAID pharmacokinetic parameters [61] . Surov et al. conducted a study on the formation of co-crystal diclofenac (DFA), diflunisal (DIF) with theophylline (THP), as well as a DFA-THP co-crystal with a DIF-THP co-crystal; the co-crystal synthesis was carried out by grinding with the addition of mixed solvent droplets (acetonitrile, methanol, and water). The synthon that formed as a hetero synthon connected by O···HN involves carboxylic acid from API and an unsaturated N atom from the imidazole ring on theophylline. The DFA-THP co-crystal provides a lower energy grid than DIF-THP because the packing consists only of dispersion energy. Co-crystal DIF-THP had a lower melting point than pure DIF, causing the co-crystal to be less stable. The small co-crystal enthalpy indicated that the energy in the hydrogen bond was proportional to the parent compound. The packaging was strengthened by the weak interactions of van der Waals forces. Based on the intrinsic dissolution results, co-crystallization increased the solubility of DFA by up to 1.3 times. In contrast, in DIF, the solubility was comparable to that of pure DIF. In this study, DIF-THP dissolution illustrated the classic concept of "spring and parachute". Apart from increasing the solubility of cocrystallization, it also increased the stability in different humidity [45] . Aitipamula et al. conducted a study of oxaprozin (OXP) co-crystal formation with the co-former 4,4 bipyridine (4.4 BP), 1,2 BPE and salts with piperazine, 2-amino-3-picoline, and anti-asthma drugs such as salbutamol [63] . OXP is included in the propionic acid group, which is used to relieve rheumatoid arthritis. It is thought to be able to inhibit urate reabsorption by selectively inhibiting glucuronidation [109] . There have been very few publications on the co-crystallization of OXP. OXP salt formation showed a decrease in intrinsic solubility and dissolution rate. Therefore, OXP co-crystal can be utilized in tablet formulation of extended-release SAL to overcome the short half-life problem and lower the frequency of SAL use. The co-crystals arrangement occurs at a stoichiometric ratio (1:0.5) with hydrogen interactions through the interactions CH···O and CH···π. The OXP-4,4 BP co-crystal had a monoclinic crystal form with one OXP molecule and half a molecule of 4.4 BP in asymmetrical units. In comparison, the OXP-1,2 BPE co-crystal produced a triclinic crystal form with one OXP molecule and half a molecule of 1,2 BPE. The co-crystal and the salt were relatively stable and not hygroscopic [63] . Drug-drug NSAID co-crystals can influence clinical practice, especially for pain management, gout therapy, and osteoarthritis. This phenomenon is related to the synergistic effect of analgesics without the potential to cause side effects, improving the physicochemical properties and clinical profile of drug release so that it is expected to reduce dosages, patient complaints related to drug use, and economic burden of treatment [110, 111] . In addition, based on the literature description of NSAID co-crystal drugs, the combination of drugs with the same BCS class can improve drugs' stability and mechanical properties, such as the flow rate and compressibility. The development of NSAID co-crystals is a complicated and lengthy process involving prediction, screening, synthesis, characterization, pre-formulation, and studies of pharmacokinetic profiles, including adsorption, distribution, metabolism, and excretion, followed by formulation, process development, preparation, declaration of an Investigational New Drug, and clinical trials [112] . The number of stages that must be passed certainly causes many significant challenges in the development of co-crystals. Choosing a co-former compatible with API is one of the challenges of co-crystal formation. Until now, the selection of co-formers was done by filtering the co-crystal by "tackles." In this approach, crystals with the best physicochemical and pharmacological properties are selected, while the selection of co-formers is performed directly by trial and error. An alternative method that is often used is the supramolecular synthons approach, where the priority of the co-former is selected based on data analysis from the CSD. In determining co-formers, several parameters should be considered. The strength of the hydrogen bonds and Hansen's solubility parameter are used to predict the theoretical solubility of drugs and co-formers. These are based on calculating partial solubility parameters with the Van Krevelen-Hoftyzer, Bagley, and Greenhalgh approach [7] . Cheney et al. conducted a study regarding the selection of co-formers in meloxicam (MLX) co-crystal formation to increase solubility and pharmacokinetics [74] . In this study, the co-former selection was carried out using the supramolecular synthons approach by analyzing the CSD to ensure the reliability of hetero-synthons or homo-synthon supramolecular formation between azole groups in MLX with carboxylic acids, primary amides, or alcohol. In the shape of the supramolecules, only hydrogen bonding interactions were considered. Based on the CSD search of 450 hits containing azole and a carboxylic acid, only 102 entries formed hetero synthon supramolecules with carboxylic acids. These results indicate that hetero-synthon supramolecules are predominantly formed, compared to homo-synthons. This result was also shown for azole-carboxylic and azole-alcohol consistently. The exception was primary amides, which contain more homo-synthon supramolecules. From these results, aspirin (ASP), which is an aromatic carboxylic acid, was selected as a co-former in co-crystal formation. However, based on the FDA approval base document, oral administration of MLX with ASP at a 1000 mg/day dose is not recommended because it can increase the AUC and Cmax of MLX, although until now, there have been no reported side effects with its use [7, 74] . Furthermore, in selecting the co-former, the safety of the co-former is the most crucial factor. The co-former established in the co-crystal formation must be safe or non-toxic in the amount required as the therapeutic drug dose. Most co-crystal development is done using co-formers that have been registered as chemical additives deemed safe for human consumption, known as GRAS. This list includes various chemicals, including aldehydes, alcohols, carboxylic acids, amides, and sweeteners. Therefore, the structural diversity and the physicochemical properties of the substances in the GRAS list provide an additional means of selecting co-formers [30] . Solubility enhancement is the main background of NSAID co-crystal development. However, some co-crystals showed no solubility improvement, i.e., niflumic acid co-crystals with sulfamethazine (SFZ) [30] . Next, ethenzamide (ET) co-crystallization with nutraceutical ingredients from hydroxybenzoic acid derivatives, namely sinapic acid (SNP), an antioxidant, antimicrobial, anti-inflammatory, and anti-cancer agent, led to lower solubility than single ET due to the poor solubility SNP. This finding indicates that the solubility of the co-former greatly influences solubility [42] . The solubility of some NSAID co-crystals depends on environmental conditions, such as pH and temperature. It was found that ET was soluble at an acidic pH of 1.2 due to chloride salt formation from protonated ET to form strong hydrogen bonds with an increase in dissolved molecules. At the same time, co-crystallization of this NSAID with 3,5 dihydroxybenzoic acid (DHBA) showed higher solubility at pH 7.4 than at pH 1.2. This shows that pH affects solubility. In addition, the conformation of a drug during the packing of the co-crystal lattice with co-former can cause molecular planarity, which leads to an increase in solubility [113] . As a ratio of the total concentration of the co-former to the whole drug in equilibrium, the crystal eutectic constant can predict the solubility at different pH [114] . Thermodynamic forms of high-energy drugs such as ionic salts and drug co-crystals with a highly soluble co-former provide the driving force to achieve drug supersaturation, called the "spring". Unfortunately, this phenomenon carries a high risk of accelerating precipitation, causing worse solubility than before. Therefore, a combination of excipients that can inhibit or slow down the precipitation or formation of "parachute" crystals is required (Figure 11 ). This approach has been applied by Guzman et al. to increase the oral absorption of celecoxib (CEL) salt co-crystal in solid preparations [115] . In this study, they conducted experiments with several excipients in the form of a surfactant mixture to obtain a parachute effect, namely the parameters of deposition time and critical micelle concentration (CMC). The deposition time was the longest, with the lowest CMC concentration. Moreover, the excipient must also be able to increase a saturated concentration of CEL in the solution. A study was carried out in vivo using beagles to determine the drug formulation with the best pharmacokinetics compared to marketed CEL, namely Celebrex. This study found that the Pluronic F127 excipient showed the best parachute effect with the inhibition of precipitation at 37 • C and a CMC concentration of 0.07 mg/mL and slowed down the dissolution or change of the CEL salt co-crystal into acidic CEL crystals. Pluronic F127 can prevent the direct wetting of CEL solids and form cohesive clots related to its flower-like form. Based on in vivo and in vitro testing results, CEL co-crystals formulation consists of vitamin E tocopherol polyethylene glycol succinate, hydroxypropyl cellulose, and Pluronic F127 and provided 100% bioavailability with faster absorption compared to Celebrex. A linear increase in the AUC with dose indicated that the formulation had a quicker onset and a lower dose. This formulation has also been tested for stability for three years. In addition, the optimal arrangement of CEL co-crystal with appropriate excipient help can increase the drug's solubility [115] . Figure 11 . The spring and parachute approach to drug saturation in solution. Adapted with permission from ref. [115] . Copyright 2007 Elsevier. Spring and parachute effects also occur in INC co-crystals with SAC as the co-former. The co-crystal induces rapid dissolution followed by sudden drug precipitation within 60 min. This phenomenon is certainly a concern in NSAID co-crystals [47] . Using appropriate excipients to inhibit co-crystal precipitation is one way of dealing with this phenomenon [112] . In addition, the formation of a co-crystal significantly affects the solubility in the construction of a supramolecular matter where the absence of a synthon with few hydrogen bonds can cause a decrease in solubility. A study of the co-crystals of meloxicam (MLX) with glutaric acid (GLU), MLX with L-malic acid, and MLX with fumaric acid showed decrease in solubility, especially for MLX-GLU co-crystal, i.e., 14% lower than that of pure MLX. In this study, the high solubility of the co-former did not translate into increased solubility of the co-crystal. The absence of NH···O = S interactions between MLX can increase solubility, but this interaction was not correlated with equilibrium Cmax changes. The poor correlation between the melting point analysis and Cmax shows that the co-crystal lattice's supramolecular arrangement was inconsistent. Intermolecular interactions between MLX molecules linked in a crystal lattice play only a limited role in crystal dissolution thermodynamics [75] . A drug's ability to penetrate biological membranes is crucial to increasing absorption and distribution to achieve its therapeutic effect. This factor is closely related to the bioavailability of a drug [112] , so that decreased permeability is a problem in the formation of NSAID co-crystals for both oral and transdermal/topical preparations. The development of NSAID co-crystals for topical preparations requires very high permeability and suitable dosage formulations. The permeability will be strongly influenced by the viscosity of the pharmaceutical preparation, as well as its pH and lipophilic/hydrophilic properties. For example, MLX-salicylic acid (SA) co-crystal in a gel was administered transdermally and showed a significant decrease in permeability, i.e., up to 5.2-fold lower than the co-crystal suspension. The formation of supersaturation due to a reduction in pH can prevent the ionization of acidic medicinal compounds, thereby increasing free drugs in the medium and increasing the permeation rate [73] . Permeability is related to lipophilic and hydrophilic properties, which depend on the density of hydrogen bonds formed on the carboxylate and hydroxy groups in the co-crystal. An increase in hydrogen bonding will increase solubility; this was observed as a rise in the ET and 3.5 DHBA co-crystal permeation rate [40] . In the other cases, γ-indomethacin-2-hydroxy-4-methylpyridine and INC-SAC co-crystals increased permeability by twofold compared to that of ordinary INC. The two co-crystals decreased the transepithelial electrical resistance value as a function of cell membrane integrity, i.e., 8.5-fold. Loss of the cell membrane monolayer (NCM460) as a barrier to drug absorption in the body can increase the drug's solubility and bioavailability and speed up the time to reach the therapeutic dose of INC [15] . Apart from INC, some co-crystal NSAID drugs have been reported to increase permeability, such as the ethenzamide-saccharine (ET-SAC) co-crystal [113] . IDR in an aqueous medium is a crucial parameter in determining the solubility of a co-crystal. IDR is the dissolution rate of a pure substance under constant temperature, pH, and surface area conditions. This parameter provides a more significant correlation with in vivo dynamic dissolution than the solubility test. IDR is a tool for evaluating drug solubility in the BCS [116] . IDR is described as the cumulative amount of dissolved per unit surface of the drug preparation plotted against time in units (unit: mg/cm 2 /min). For example, co-crystallization of IBU with NIC could increase the dissolution rate by up to 2.5 times that of the single IBU [59] . The tramadol-celecoxib co-crystal (CTC) also increased dissolution rate three times more than CEL, which can accelerate absorption and increase bioavailability. The maximum concentration of CEL in CTC increased, compared to CEL alone [79] . Different co-crystal preparation methods result in various particle sizes and shape distributions, affecting the IDR. Based on previous research, co-crystallization between PCM and caffeine (CAF) produces co-crystals A, B, C, D, E, which showed increased IDR values 1.72, 1.88, 2.42, 2.19, and 2.84 times higher than that of IDR from PCM (5.06 mg/cm/min) [82] . The IDR can be comparable to or even less than the dissolution rate of a single NSAID drug, such as the IBU-NIC and flurbiprofen (FLU)-NIC co-crystals, which prolonged IDR around eight times lower than IBU and five times lower than FLU [117] . The pH microenvironment plays a vital role in the solubility and dissolution of the cocrystal. This is shown by the initial pH (pHint) relationship representing the bulk pH, with equilibrium pH representing the microenvironment pH ( Figure 12 ). In a co-crystal, pHint will reach equilibrium at the eutectic time, in which the solution is highly saturated with the co-crystal and drug. In contrast to drugs, where pHint = bulk pH, for co-crystals pHint