Translatability of in vitro potency to clinical efficacious exposure: A retrospective analysis of FDA‐approved targeted small molecule oncology drugs

Abstract

In vitro potency is one of the important parameters representing efficacy potential of drugs and commonly used as benchmark of efficacious exposure at early clinical development. There are limited numbers of studies which systematically investigate on how predictive in vitro potency is to estimate therapeutic drug exposure, especially those focusing on targeted anticancer agents despite the recent increase in approvals. This study aims to fill in such knowledge gaps. A total of 87 small molecule targeted drugs approved for oncology indication between 2001 and 2020 by the US Food and Drug Administration (FDA) were identified; relevant preclinical and clinical data were extracted from the public domain. Relationships between the in vitro potency and the therapeutic dose or exposure (unbound average drug concentration [C _u,av] as the primary exposure metrics) were assessed by descriptive analyses. The Spearman's rank correlation test showed slightly better correlation of the C _u,av (ρ = 0.232, p = 0.041) rather than the daily dose (ρ = 0.186, p = 0.096) with the in vitro potency. Better correlation was observed for the drugs for hematologic malignancies compared with those for solid tumors (root mean square error: 140 [n = 28] versus 297 [n = 59]). The present study shows that in vitro potency is predictive to estimate the therapeutic drug exposure to some extent, whereas the general trend of overexposure was observed. The results suggested that in vitro potency alone is not sufficient and robust enough to estimate the clinically efficacious exposure of molecularly targeted small molecule oncology drugs. The totality of data, including both nonclinical and clinical, needs to be considered for dose optimization.

Study Highlights.

• WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC?

In vitro potency is one of the important parameters representing efficacy potential of drugs and is commonly used as a benchmark to predict clinically efficacious exposure levels. However, there is limited knowledge regarding the translatability of in vitro potency to estimate the therapeutic drug exposure, especially those focusing on molecularly targeted small molecule oncology drugs despite the recent increase in approvals.

• WHAT QUESTION DID THIS STUDY ADDRESS?

This study addressed on how predictive in vitro potency is to estimate therapeutic drug exposure for molecularly targeted small molecule oncology drugs by systematically analyzing the relationships between in vitro potency and therapeutic drug exposure of the targeted drugs approved by the US Food and Drug Administration.

• WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE?

The results suggested that in vitro potency is predictive to estimate the therapeutic drug exposure to some extent and may be informative during the decision making in the drug development, whereas general trend of overexposure suggested it is not sufficient and robust enough to justify dose optimization of molecularly targeted small molecule oncology drugs.

• HOW MIGHT THIS CHANGE CLINICAL PHARMACOLOGY OR TRANSLATIONAL SCIENCE?

This might provide precautions of using in vitro potency alone as a benchmark of drug efficacy in the process of molecularly targeted small molecule oncology drug development. In addition, this might facilitate translational research from preclinical to clinical in oncology drug development and shift in dose optimization paradigm for targeted drugs.

INTRODUCTION

The drug development paradigm in an oncology field has drastically evolved in the past few decades. Historically, chemotherapies with cytotoxic agents which cause cell killing not only for tumor cells but also for healthy normal cells had been the mainstream of drug therapy in cancer treatment for a long time. However, since the approval of the first molecularly targeted small molecule anticancer drug, imatinib, by the US Food and Drug Administration (FDA) in 2001, ¹ the trend of oncology drug discovery and development has shifted from broad‐spectrum cytotoxic drugs to targeted drugs, and the number of approvals of molecularly targeted small molecule oncology drugs has significantly increased along with advances in the understanding of cancer disease biology, its molecular drivers, pathophysiology, and human pharmacology. ² , ³

Therapeutic concept of molecularly targeted small molecule oncology drugs is totally different from that of cytotoxic chemotherapies, which usually assume high correlation between toxicity/safety and efficacy due to its nontumor‐specific mechanism of action, and whose dose selections are primarily driven by the assumption that the maximum tolerated dose (MTD) should be given in order to achieve as much high cytotoxic effect on tumor cells as possible. ⁴ , ⁵ , ⁶ Molecularly targeted drugs are designed and expected to demonstrate their pharmacological activities via engagements to its target molecules which expressed on or exhibit abnormal functions in tumor cells specifically. Therefore, the in vitro potency of compound for its target is considered to be one of the most important features that represents therapeutic potential of the drug and may, at least partly, explain the target efficacious exposure and inform recommended dose selection of drug. It is usually assessed as a part of various preclinical pharmacology studies carried out to investigate the mechanism of action and potential efficacy of the agent which include primary pharmacodynamic assessments in cell‐free biochemical assays and phosphorylation assays, cell growth inhibitory effects in relevant in vitro cell lines (i.e., cell‐based assays), and in vivo antitumor activity in mouse xenograft models. ⁷

There are limited numbers of studies which systematically investigate the translatability/predictability of in vitro potency to estimate the therapeutic drug exposure so far. One recent report tried to address the general relationship between clinical unbound therapeutic concentrations and in vitro potency, on how they may differ depending on therapeutic indication, mode of action, receptor type, target localization and function, and potential drug discovery and development implications through the analysis of 164 marketed small molecule drugs. ⁸ However, according to the Anatomical Therapeutic Chemical (ATC) Classification System, only six anticancer agents categorized as ATC code L (i.e., antineoplastic and immunomodulating agents) out of 164 were included in the analysis. Moreover, both traditional cytotoxic agents and targeted drugs were mixed in those six anticancer agents (etoposide, gefitinib, imatinib, letrozole, topotecan, and vorinostat), which further made limitations on the evaluation and the interpretation of the results in this specific category.

The present study was conducted (i) to address translatability of the in vitro potency to estimate the clinically efficacious drug exposure, and (ii) to exploratory assess potential influential factors that may affect the relationships between the in vitro potency and the therapeutic drug exposure, especially focusing on molecularly targeted small molecule oncology drugs. A retrospective analysis of 87 FDA‐approved targeted small molecule oncology drugs was conducted to explore the correlations between the in vitro potency (biochemical assays readout) and the therapeutic drug exposure in humans.

METHODS

Selection of drugs

Molecularly targeted small molecule oncology drugs, approved between January 2001 (the approval year of imatinib, the first molecularly targeted anticancer drug) and December 2020 by the FDA, were identified from the list of “Compilation of CDER New Molecular Entity (NME) Drug and New Biologic Approvals.” The list is publicly available at the FDA's website (https://www.fda.gov/drugs/drug‐approvals‐and‐databases/compilation‐cder‐new‐molecular‐entity‐nme‐drug‐and‐new‐biologic‐approvals). The drugs were selected based on the following three criteria: (i) it is approved for oncology indication, (ii) its therapeutic modality can be classified as small molecule, and (iii) it has distinct target molecule(s) to demonstrate its expected pharmacological activity. After identifying the drugs of interest for the present study, the FDA Summary Basis of Approval (SBA) and the US label were obtained on the Drugs@FDA website (https://www.accessdata.fda.gov/scripts/cder/daf/) and reviewed to extract relevant preclinical and clinical data for the analyses, as detailed in the following sections. All of these data are available in Table S1.

In vitro potency

The target molecule(s) and the in vitro potency against it for each small molecule oncology drug, represented as the half maximal inhibitory concentration (IC₅₀) from biochemical assay, were extracted from the FDA pharmacology SBAs. Although in vitro potency data from cell‐based assay were also available for most of the drugs, the IC₅₀ from biochemical assay were used as the metrics of in vitro potency in the present study because it was thought to represent a pure potency of each drug to its target molecule. In case of cell‐based assay, even if intended to evaluate drug potency against the same target molecule, not only the drug potency but some other factors, such as drug sensitivity to a specific cell line, permeability of drug toward cell membrane (especially for drugs which have intracellular target), can differ from cell line to cell line used in each in vitro experiment and are likely to confound in vitro potency readout. Thus, side by side comparison of the cell‐based in vitro potency values reported for each drug was considered not appropriate.

The in vitro potency associated with the target molecule, which is expected to lead to antitumor activity, was collected for each drug. If a drug had more than one reported pharmacologically relevant target molecules (e.g., tropomyosin receptor kinases [TRK] A, B, and C, ROS proto‐oncogene1 [ROS1], and anaplastic lymphoma kinase [ALK] proteins for entrectinib), the geometric mean of the reported IC₅₀ values toward all relevant target molecules was used because pharmacological effect given by such drug in the in vivo situation is likely to be shown as mixed effects against all relevant targets rather than only one of them.

Therapeutic dose and clinical exposures

Approved dosing regimen (including the therapeutic dose and dosing interval) for the primary indication, the geometric mean steady‐state area under the plasma concentration‐time curve during the dosing interval (AUC_tau) at the approved dosing regimen, and fraction of drug unbound to plasma proteins determined in vitro (f _u) were extracted from the US label as well as from the clinical pharmacology SBAs. This study focused on the primary indication (i.e., the first approved indication) of each drug because it was supposed to be most relevant to the expected mechanism of action; the clinical development for the primary indication was expected to be supported by the preclinical data most, rather than the subsequent indications whose development could be supported by the existing clinical data.

The daily dose (i.e., dose per 24 h) was derived by following Equation 1:

$$\text{Daily Dose}=\text{Therapeutic Dose}\times \frac{24}{\mathrm{tau}}$$ (1)

where tau is the dosing interval in an hourly unit. The majority of the approved molecularly targeted small molecule oncology drugs were intended for oral administration, and, thus, the therapeutic doses were defined in fixed per body dose basis (i.e., mg/body) in general. However, some drugs had the therapeutic doses defined in per body surface area basis (i.e., mg/m²). In this case, the average adult body surface area of 1.73 m² was assumed. ⁹

The total and unbound average plasma concentration at steady‐state (C _tot,av and C _u,av, respectively) for each drug was derived by following Equations 2 and 3:

$$C_{\mathrm{tot},\mathrm{av}}=\frac{{\mathrm{AUC}}_{\mathrm{tau}}}{\mathrm{tau}}$$ (2)

$$C_{\mathrm{u},\mathrm{av}}=f_{\mathrm{u}}\times C_{\mathrm{tot},\mathrm{av}}$$ (3)

To derive C _tot,av and C _u,av in a molar unit (e.g., nM), AUC_tau was corrected by the molecular weight of each drug if it was reported based on a mass per volume unit (e.g., ng•h/mL). Average plasma concentration at steady‐state was used as the primary exposure metrics in this study rather than trough plasma concentration at steady state (C _trough), the drug concentration at the end of the dosing interval just prior to the next dose, because C _trough was not available either in the US labels or the clinical pharmacology SBAs in many cases (available only for 36% of the drugs in this study). On the other hand, AUC_tau and dosing interval could generally be obtained from these source domains and C _tot,av or C _u,av could be easily derived.

Other target or drug‐related features

To assess its potential influence on relationships between drugs' in vitro potency and clinical exposure, information about localization of the target molecule(s) and presence or absence of pharmacologically active metabolites for each drug were also collected. The localization of each target molecule was searched from a publicly accessible database, The Human Protein Atlas (https://www.proteinatlas.org/), in which comprehensive protein expression profile data in human are summarized; localization‐type was categorized as “Intracellular,” “Membrane,” or “Membrane and Intracellular.” The presence or absence of pharmacologically active metabolites of each drug in humans was extracted from the FDA clinical pharmacology SBAs. Classification of the pharmacologically active metabolite for each drug was categorized as “No” if its presence is not reported in the clinical pharmacology SBAs, or if it is reported but its exposure level was less than 10% of total drug exposure ¹⁰ ; otherwise, it was categorized as “Yes.”

Graphical and statistical analyses

Correlation between in vitro potency and daily dose, C _tot,av or C _u,av was assessed descriptively with scatter plots. To investigate those clinical pharmacokinetic‐related metrics relation to the in vitro potency semiquantitatively, the Spearman's rank correlation test was performed. The correlation plot of in vitro potency and C _u,av was also divided by a type of cancer (solid tumor vs. hematologic malignancy) of which the drugs were intended to be used for treatment to see if there were any differences in the overall trend of the relationships. In addition, root mean square error (RMSE) was calculated by following Equation 4 where appropriate and compared to evaluate differences of correlation trends in different settings (e.g., C _tot,av vs. C _u,av, solid tumor vs. hematologic malignancy) quantitatively:

$$\text{RMSE}=\sqrt{\sum _{i=1}^n\frac{{\left({\widehat{y}}_i-y_i\right)}^2}{n}}$$ (4)

where ŷ _1…n are the C _tot,av or C _u,av of drugs 1 to n, and y _1…n are the in vitro potency of drugs 1 to n, respectively.

Potential influence of target localizations and presence or absence of pharmacologically active metabolites were investigated descriptively by comparing the distribution of the ratio of C _u,av/in vitro potency in each subgroup with violin plots. In addition, the Wilcoxon rank‐sum test was carried out to compare the distributions of the C _u,av/in vitro potency ratios between the two subgroups where applicable.

Drugs were excluded from the above analyses if any of the data needed (i.e., in vitro potency, daily dose, C _tot,av, or C _u,av) were missing. R software version 4.0.3 (The R Foundation for Statistical Computing, Vienna, Austria) was used for general scripting, data management, graphical analyses, and statistical testing.

RESULTS

Disposition of selected drugs

A total of 657 new molecular entities (NMEs) were approved by the FDA between January 2001 and December 2020 according to the FDA's list of “Compilation of CDER New Molecular Entity (NME) Drug and New Biologic Approvals.” Based on the drug selection criteria described in Methods, 87 molecularly targeted small molecule oncology drugs were identified for the present study (Figure 1). The majority of the drugs were approved for treatment of solid tumors (n = 59 [68%] for solid tumor indications vs. n = 28 [32%] for hematologic malignancy indications; Figure 2a). Within the drugs approved for treatment of solid tumors, more than four drugs were approved for treatment of non‐small cell lung cancer (n = 14), breast cancer (n = 9), prostate cancer (n = 7), melanoma (n = 6), and renal cell carcinoma (n = 5). Similarly, within the drugs approved for treatment of hematologic malignancies, more than four drugs were approved for treatment of chronic myeloid leukemia (n = 6), acute myeloid leukemia (n = 5), and multiple myeloma (n = 5; Figure 2b). Most of the drugs were for oral use (n = 77), whereas a few of them were intended for other routes of administration (n = 6 for intravenous use, n = 2 for subcutaneous use, and n = 2 for intramuscular use). The number of drugs with its target localized at the intracellular space was 45 (52%), those with its target localized on cell membrane was 37 (42%), and those with its target localized both on cell membrane and at intracellular space was five (6%). The presence of pharmacologically active metabolites was reported for 30 (35%) drugs, whereas other 57 (65%) drugs were reported not to have clinically meaningful active metabolites in humans.

FIGURE 1.

Diagram of the selection of molecularly targeted small molecule oncology drugs approved by the US Food and Drug Administration (FDA) between January 2001 and December 2020. NME, new molecular entity.

FIGURE 2.

Type of cancer (a) and primary indications (b) for which the selected drugs were approved. ALL, acute lymphocytic leukemia; AML, acute myeloid leukemia; BC, breast cancer; BCC, basal cell carcinoma; CCA, cholangiocarcinoma; CLL, chronic lymphocytic leukemia; CML, chronic myeloid leukemia; CRC, colorectal cancer; CTCL, cutaneous T‐cell lymphoma; ES, epithelioid sarcoma; FL, follicular lymphoma; GIST, gastrointestinal stromal tumor; MCL, mantle cell lymphoma; MF, myelofibrosis; MM, multiple myeloma; MTC, medullary thyroid carcinoma; NF, neurofibromas; NSCLC, non‐small cell lung cancer; NTRK, neurotrophic tyrosine receptor kinase; OC, ovarian cancer; PC, prostate cancer; PTCL, peripheral T‐cell lymphoma; RCC, renal cell carcinoma; RR‐DTC, radioiodine‐refractory differentiated thyroid cancer; ST, solid tumor; TGCT, tenosynovial giant cell tumors; UC, urothelial cancer.

In these 87 drugs, nine drugs had missing data for either in vitro potency (n = 5; gilteritinib, nilotinib, romidepsin, temsirolimus, and vismodegib) or AUC_tau (and corresponding C _tot,av and C _u,av; n = 3; belinostat, bortezomib, and midostaurin) or both (n = 1; omacetaxine). These drugs were excluded from the subsequent analyses. Distributions of the in vitro potency, the daily dose, the C _tot,av, and the C _u,av showed wide ranges, with the median (min–max) values of 5.21 (0.1–294) nM, 180 (0.571–1920) mg, 620 (5.19–102,000) nM, and 21.3 (0.0784–1310) nM, respectively. The ratio of C _u,av/in vitro potency distributed in a range of 0.0183–987 with the median ratio of 3.32.

Correlation between in vitro potency and therapeutic dose or clinical exposure

Relationships between the in vitro potency (the IC₅₀ from biochemical assay) and the pharmacokinetic‐related metrics, either of the daily dose, the C _tot,av, or the C _u,av, for each drug are presented in Figure 3. Generally, all the pharmacokinetic‐related metrics showed a trend of positive correlation with the in vitro potency. The Spearman's rank correlation coefficient for correlation with the in vitro potency was highest for the C _tot,av (ρ = 0.313, p = 0.0052), followed by the C _u,av (ρ = 0.232, p = 0.041) and the daily dose (ρ = 0.186, p = 0.096). The RMSE value calculated from the C _u,av and the in vitro potency of drugs was much lower compared with that calculated from the C _tot,av and the in vitro potency (264 vs. 16,800). When comparing the in vitro potency and the C_tot,av, all the drugs except for only one drug (panobinostat) showed the C _tot,av greater than the in vitro potency values. When comparing the in vitro potency and the C _u,av, most of the drugs (n = 56, 64%) showed the C _u,av higher than the in vitro potency with many of them exceeding 10‐fold separation (n = 27, 31%). However, notably, nine drugs showed the C _u,av less than one‐third of the in vitro potency values, of which seven drugs showed the C _u,av less than one‐tenth of the in vitro potency values (Table 1).

FIGURE 3.

Correlations between the in vitro potency (the half maximal inhibitory concentration [IC₅₀] from biochemical assay) and the daily dose, the C _tot,av, or the C _u,av. Each point represents a drug. Blue solid line and gray shaded area depict a linear regression line and 90% confidence interval. Black solid line is the line of unity and black dotted lines represent 10‐fold range of the in vitro potency versus the C _tot,av or the C _u,av. C _tot,av, total average plasma concentration at steady‐state; C _u,av, unbound average plasma concentration at steady‐state.

TABLE 1.

Drugs with moderate to large underexposure discrepancy between the C _u,av and the in vitro potency values (the C _u,av <1/3 of the in vitro potency values).

Drug name	Type of cancer	Indication	Target protein	C u,av/in vitro potency ratio
Neratinib	Solid	Breast cancer	EGFR/HER2/HER4	0.0183
Fulvestrant	Solid	Breast cancer	Estrogen Receptor	0.0206
Panobinostat	Heme	Multiple myeloma	HDAC	0.0358
Ponatinib	Heme	Chronic myeloid leukemia	Multiple kinases	0.0364
Ixazomib	Heme	Multiple myeloma	20S proteasome	0.0421
Axitinib	Solid	Renal cell carcinoma	VEGFR	0.0885
Ripretinib	Solid	Gastrointestinal stromal tumor	KIT/PDGFRA	0.0975
Abiraterone	Solid	Prostate cancer	CYP17	0.119
Carfilzomib	Heme	Multiple myeloma	20S proteasome	0.150

Abbreviation: C _u,av, unbound average drug concentration.

Influence of other drug or target‐related features

Relationships between the in vitro potency and the C _u,av divided by the type of cancers (i.e., solid tumors vs. hematologic malignancies) of which the drugs were approved for treatment are shown in Figure 4. Overall, the drugs approved for treatment of hematologic malignancies had closer in vitro potency and C _u,av values (plots scattered around the line of unity x = y), whereas the drugs approved for treatment of solid tumors showed the more general trend of overexposure (the C _u,av greater than the in vitro potency). The RMSE values calculated from the C _u,av and the in vitro potency of drugs were lower for the drugs approved for treatment of hematologic malignancies compared with those approved for treatment of solid tumors (140 vs. 297).

FIGURE 4.

Correlations between the in vitro potency (the half maximal inhibitory concentration [IC₅₀] from biochemical assay) and the C _u,av by the type of cancers. Each point represents a drug. Blue solid line and gray shaded area depict a linear regression line and 90% confidence interval. Black solid line is the line of unity and black dotted lines represent 10‐fold range of the in vitro potency vs the C _u,av. C _u,av, unbound average plasma concentration at steady‐state.

The distribution of the C _u,av/in vitro potency ratios in the subgroups divided by target localizations (“Intracellular,” “Membrane,” or “Membrane and Intracellular”) or presence of pharmacologically active metabolites (No/Yes) are shown in Figure 5. Overall, the violin plots were largely overlapped and there was no apparent difference in the distribution of the C _u,av/in vitro potency ratios across the different subgroups divided by target localizations. On the other hand, there was a slight tendency of the lower C _u,av/in vitro potency ratios in the subgroup of drugs with pharmacologically active metabolites compared with those without, and the difference in the distribution was statistically confirmed by the Wilcoxon rank‐sum test (p = 0.0313).

FIGURE 5.

Distribution of the C _u,av/in vitro potency ratios in the subgroups divided by target localizations “Intracellular (I),” “Membrane (M),” or “Membrane and Intracellular (M/I)” or presence of active metabolites “No” or “Yes”. *p < 0.05, the Wilcoxon rank‐sum test. C _u,av, unbound average plasma concentration at steady‐state.

DISCUSSION

Since the approval of imatinib by the FDA in 2001, the trend of oncology drug discovery and development has shifted from broad‐spectrum cytotoxic drugs to targeted drugs, and a lot of molecularly targeted small molecule oncology drugs have been developed and approved to date. Due to the difference in its mechanism of action compared with historical cytotoxic agents, necessity of a paradigm shift in the recommended clinical dose selection from a traditional MTD‐based (i.e., toxicity‐driven) dosing strategies to alternative ones has been discussed in several reports. ¹¹ , ¹² , ¹³ Furthermore, as represented by the launch of the FDA's Project Optimus ¹⁴ and related draft guidance, ¹⁵ how to estimate clinically efficacious exposure and to inform the dose selection for novel oncology drugs is one of the most important topics of interest in this area. In the present study, correlation between in vitro potency and clinical therapeutic dose/exposure for 87 molecularly targeted small molecule oncology drugs that have been approved by the FDA was systematically investigated, and translatability/predictability of in vitro potency to clinically efficacious exposure was assessed to shed light on this topic.

The distribution of the ratios of C _u,av/in vitro potency for the studied 87 molecularly targeted small molecule oncology drugs (median: 3.32, 64% were greater than unity) tends to be larger compared with the previous report, which assessed 164 marketed small molecule drugs across various therapeutic areas (median: 0.32, ~70% were less than unity). ⁸ This may be due to a difference in the scope of the therapeutic area in each study; the present study focuses on the oncology area in which traditional MTD‐based approach (i.e., the higher exposure, the better efficacy) has been mostly used for dose selection.

Correlation analysis between the in vitro potency and the pharmacokinetic‐related metrics showed a general trend of positive correlation for all of the daily dose, the C _tot,av, or the C _u,av with the in vitro potency (Figure 3). It is reasonable that the drug exposures (C _tot,av and C _u,av), which are the more direct “driver” of the pharmacological activity of molecularly targeted small molecule oncology drugs in vivo, showed better correlation than the daily dose, which is just an input index of drugs to human bodies. Notably, the C _tot,av seems to explain the rank‐order of the drugs' potency better than the C _u,av as shown by the higher Spearman's rank correlation coefficient (ρ = 0.313 for the C _tot,av vs. ρ = 0.232 for the C _u,av), but the RMSE values (16,800 for the C _tot,av vs. 264 for the C _u,av) suggested that the unbound plasma drug concentrations are more quantitatively relevant to the drugs' potency. This is consistent with the generally accepted concept that the unbound (not the bound nor the total) plasma drug concentrations drive the pharmacological effect of small molecule drugs. ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ The f _u values determined in vitro were quite small for many of the studied drugs (53 out of a total of 87 drugs in this study had f _u ≤0.05), and in vitro experimental limitations to determine the f _u accurately may be confounding the rank‐order explainability of the C _u,av compared with the C _tot,av.

Many of the molecularly targeted drugs developed and approved for solid tumor indications had the C _u,av greater than the in vitro potency, and in comparison, those for hematologic malignancies tend to have the C _u,av closer to the in vitro potency values (Figure 4). This could be explained by a difference in drug delivery to the actual site of action. In treatment of solid tumors, the drugs need to be delivered to the tumor site in the target organ after entering the systemic circulation in order to exert its pharmacological activity. Therefore, there could be a discrepancy between the unbound drug concentrations in plasma and near the target protein at the tumor site. On the other hand, drugs in the systemic circulation are supposed to be at or very close to the site of action for hematologic malignancies, which could explain smaller separation between the C _u,av and the in vitro potency values.

For the nine drugs which showed the C _u,av less than one‐third of the in vitro potency values, of which seven drugs showed the C _u,av less than one‐tenth of the in vitro potency values (Table 1), there was no particular trend in terms of the type of cancers (5 drugs were for solid tumor indications and 4 drugs were for hematologic malignancy indications). Taking a closer look at actual indication of use, three drugs were for multiple myeloma treatment (out of a total of 4; 75%), one drug was for gastrointestinal stromal tumor treatment (out of a total of 3; 33%), one drug was for chronic myeloid leukemia treatment (out of a total of 4; 25%), two drugs were for breast cancer treatment (out of a total of 9; 22%), one drug was for renal cell carcinoma treatment (out of a total of 6; 17%), and one drug was for prostate cancer treatment (out of a total of 7; 14%). Although the C _u,av was closer to the in vitro potency values for the drugs for hematologic malignancy indications in general, three out of a total of four drugs (75%) for multiple myeloma treatment are found to be largely underexposed (i.e., the C _u,av less than the in vitro potency). Interestingly, two of those three drugs (carfilzomib and ixazomib) have 20S proteasome as a target molecule and these are the only drugs targeting 20S proteasome in the evaluable drugs in the present study. There might be target‐specific limitations in the translatability of in vitro potency to clinically efficacious exposure.

Although it could be assumed that the drugs with membrane targets may need less C _u,av than the drugs with intracellular targets because the former drugs can directly access to the targets without crossing cell membrane, there was no apparent trend in the distribution of the C _u,av/in vitro potency ratios in the subgroups divided by target localizations. Even if the target molecule itself is reported to be mainly localized on the membrane, it is likely to exist as transmembrane form, and some of the drugs should have its target for actual site of action at the intracellular domain. In addition, if active transport mechanism of drugs exists, intracellular C _u,av can possibly be greater than extracellular C _u,av. These may mask the true underlying trend and make the results and the interpretation unintuitive. Further methodological evolutions to assess intracellular drug exposure and deeper understanding of the relationship between free drug concentrations in plasma and within tissues is needed. ²⁰ , ²¹ , ²² , ²³

The drugs with reported pharmacologically active metabolites tended to show the lower C _u,av/in vitro potency ratios compared with those without such metabolites (Figure 5). This result suggested that the presence of active metabolites may support exertion of the intended pharmacological activity or cause additional toxicities on top of those by the parent compound, which lead to lower the clinical exposure of the parent compound. The presence of active metabolites needs to be taken into account and would cause additional complexity on the recommended dose selection for the targeted drugs given that parent compound and active metabolites usually have different pharmacokinetic properties as well as in vitro potencies. ²⁴

In conclusion, the present study attempted to investigate on the translatability of in vitro potency to therapeutic exposure for molecularly targeted small molecule oncology drugs approved by the FDA between 2001 and 2020. The results suggested that in vitro potency is predictive to estimate the therapeutic drug exposure to some extent and may be informative during the decision making in the drug development. General trend of overexposure was observed and lack of insights on how much clinical exposure should exceed preclinical target exposures is likely to limit using in vitro potency alone as a benchmark to guide dose selections. In vitro potency based estimation of the clinically efficacious exposure is not sufficient and robust enough to justify dose optimization of molecularly targeted small molecule oncology drugs and the totality of data, including both nonclinical and clinical, needs to be considered.

AUTHOR CONTRIBUTIONS

N.K. and K.I. designed the research, performed the research, and wrote the manuscript. N.K. analyzed the data.

FUNDING INFORMATION

No funding was received for this work.

CONFLICT OF INTEREST STATEMENT

N.K. is a current employee of Chugai Pharmaceutical Co., Ltd. K.I. declared no competing interests for this work.

Supporting information

Table S1

Kotani N, Ito K. Translatability of in vitro potency to clinical efficacious exposure: A retrospective analysis of FDA‐approved targeted small molecule oncology drugs. Clin Transl Sci. 2023;16:1359‐1368. doi: 10.1111/cts.13532