4. Discussion

A review published in 202022 had reported nine PopPK models of posaconazole (one could not be found online, and the full text was not available even after contacting the author). Our review incorporated five new published models into the scope of examination. We focused only on the PopPK of posaconazole for the first time, providing a simulation of posaconazole exposure at different covariates levels and dosing regimens of 13 published models.
Without limiting the population, only three of our included studies considered the pediatric population as the primary study population29,32,33. During the literature screening, there were few PK or clinical reports of posaconazole in the pediatric population, which may be related to the limited use of posaconazole in pediatrics. Posaconazole has not been approved for use in children under 13 years of age. Nevertheless, there have been some cases of posaconazole being used off-label for the prevention of high-risk IFI in children ≤12 years old36. This is not only due to the satisfactory efficacy and safety of posaconazole in adults37,38, but also because posaconazole is more effective than other antifungal agents such as fluconazole and itraconazole in pediatric patients with hematologic malignancies39-42. Plasma concentrations of posaconazole are highly variable in the younger pediatric population43,44, which may lead to large fluctuations in efficacy and safety. In pediatric patients treated with posaconazole, TDM is necessary to ensure that the required drug exposure is achieved and to minimize the occurrence of adverse events.
In this review, the structural model appears to be linked to the route of administration, as demonstrated by the fact that the two studies30,34 involving intravenous administration used two-compartment models, while the studies of oral administration used one-compartment models. Since most studies used sparse sampling lacking absorption phase data or fixed ka to a specific value according to the literature, inaccurate estimation of kamight have affected the judgment of structural models. In addition, two models with absorption delays18,33 may have obscured the initial distribution pattern30.
The sample size, evaluation method, inclusion and exclusion criteria of covariates, pathological status, and concomitant medications were different in different studies, which may lead to differences in the influence of covariates in each study.
Diarrhea, a common symptom in patients with graft-versus-host disease (GVHD), critically ill patients, and patients after receiving chemotherapy, is associated with a significant decrease in F45,46. Nearly half of the studies in our review retained diarrhea in the final model. The F of posaconazole was reduced by 59% and 45% in the adult models M123 and M518, respectively. In pediatric study models, M829 and M1233, it was reduced by 33% for both. Additionally, the presence of diarrhea in M224 and M325 increased V and CL by a factor of 1.5. M728 examined but did not retain diarrhea in the final model. Unlike the six studies mentioned above, the formulation of posaconazole used in M728 was a delayed-release tablet rather than an oral suspension. Diarrhea was a risk factor for sub-therapeutic concentration of posaconazole in patients using tablets, but there was a decreasing trend observed in this effect47,48. Metoclopramide, which was retained in M518, similar to the diarrhea limited the absorption and altered the exposure of posaconazole by increasing gastrointestinal motility.
The use of PPI was considered an important covariate examined in six models18,24,25,27,29,32,33, of which were retained except M1132. The ultimate effect of the use PPI in these models was manifested by reduced plasma exposure with the form of raising V or CL, or decreasing F, which was consistent with the results reported in other articles15,17,49. PPI can effectively prevent stress mucositis in critically ill patients50,51 by inhibiting the secretion of gastric acid and increasing the pH of gastric juice. However, for posaconazole, a weakly alkaline drug, its solubility and F may be altered by the concomitant use of PPI13. M829 found that PPI limited posaconazole absorption to a greater extent than H2 receptor antagonists. This may be due to the stronger and longer-lasting acid inhibitory effect of PPI than H2receptor antagonists52.
Demographic characteristics such as weight, age, and sex were also examined. The influence of body weight on V, CL, and F of posaconazole are described in several models25,28-30. The high lipophilicity of posaconazole may be responsible for extensive lipid tissue distribution53, which may account for the greater V in individuals with high body weight. Sex and age were tested in most studies but were retained only in M1031 and M123, respectively. M1031 showed lower CL in women than in men, consistent with the finding that males were associated with reduced posaconazole trough concentrations as mentioned in three reports47,54,55. On the contrary, some studies have found that men have higher plasma exposure than women (P = 0.028)56,57. Jia et al.54speculated that differences in sex hormones and fat content between men and women contributed to the varied PK of posaconazole. Despite the fact that age was considered to be relevant to the decrease of V in M123, the effect of age on posaconazole concentration was not noticeable in our simulations, which may be explained by the low plasma exposure caused by the large V in M1.
Some studies have also considered the effect of biochemical indicators on the PK of posaconazole. Posaconazole has a plasma protein binding rate of 98% and is primarily bound to albumin58. Restricted transmembrane transport caused by protein binding results in a reduction in metabolism and excretion and an elevation of plasma concentrations, which fits with the findings of M1031. However, this study did not find a relationship between albumin and PK parameters, indicating that the CL/F of posaconazole may be influenced by other plasma-binding proteins such as lipoprotein59,60 and C-reactive protein (CRP)54. M224 found that posaconazole exposure decreased with the baseline bilirubin ≥ 2 × the upper limit of normal (ULN) or GGT ≥ 2 × ULN24. This may be an indirect effect caused by metabolic disorders due to liver impairment, although liver function is not an absolute condition for changes in bilirubin61 and GGT62-65levels. Other biochemical markers such as alanine aminotransferase (ALT), aspartate aminotransferase (AST), and alkaline phosphatase (ALK) were also tested in some models but were not retained.
The effect of concomitant medications on posaconazole exposure was mainly reflected in M518. Phenytoin and rifampin presented a remarkable effect on CL/F (621% increase). This effect may arise from enzymatic interactions; phenytoin and rifampin, inducers of the UGT enzyme66,67, increase the metabolism of posaconazole, which is metabolized by UGT1A4 by approximately 17%22. These two drugs were also tested by M829 and M1132 but were not retained, possibly because the populations in both studies were pediatric with immature expression of drug-metabolizing enzymes or because of the low proportion of patients with concomitant use of these two drugs. Fosamprenavir also increased CL/F, although this effect was much less than that of phenytoin and rifampin. M518reported that nutritional supplements increased the F of posaconazole by 129%, in agreement with the findings of published studies15,68,69. PK studies have demonstrated that food, especially a high-fat diet, can greatly increase the rate and extent of posaconazole absorption70-72. However, for patients with eating disorders due to severe IFI, liquid nutritional supplements are often used as a substitute of food for enteral nutrition69. Furthermore, M325revealed a 0.6-fold decrease in V as a result of the co-administration of chemotherapy. In conclusion, TDM is advisable when used in combination with drugs that may alter the PK of posaconazole.
Regardless of the covariate or dose simulations, there were observable differences in posaconazole steady-state concentrations between models, even at the same dose. Such differences may derive from variation in the race, age, or disease state of the population, the formulation of posaconazole, and the assay conditions of the plasma samples among studies. Nevertheless, the pattern of covariate or dose effects on the exposure of posaconazole was mostly consistent. According to the simulated PK profile, posaconazole tablets and intravenous formulations showed higher concentrations than oral suspensions, which was consistent with the reported finding73. This might be because delayed-release tablets with drug-polymer combinations prevent drug recrystallization in the intestinal fluid and therefore exhibit higher F than suspensions74.
Since only a small number of studies used non-parametric modeling methods75,76, we only retained studies using parametric modeling methods, which also ensured the comparability among models. Further discussion is needed if more non-parametric studies are conducted in the future. The other limitation is that the models in this review were evaluated using internal data. Thus, the good predictive performance of the models is only reflected in their own centers and is difficult to apply when extrapolated to other centers. A more rigorous external evaluation of these models is recommended to verify their predictive performance and robustness after extrapolation to other scenarios.