Introduction

For many decades, drug–drug interactions (DDI) have formed a major clinically important problem of drug treatment; cytochrome P450 (CYP) enzymes being the most important phase I xenobiotic-metabolizing enzymes involved in the DDIs (see Pelkonen et al 1998; 2008 and Hakkola et al. this issue).

For us, it started really with cimetidine. One of the earliest cases of CYP-associated DDIs was cimetidine, at the time a novel histamine H2 receptor inhibitor developed for gastroesophageal reflux disease and ulcers. Hepatic microsomal studies were already at that time applied for the study of CYP enzymes and seminal studies in the 1979 and 1980 indicated that cimetidine inhibited the metabolism of some CYP-catalyzed activities in human and animal liver preparations as well as in vivo in humans and animals (Puurunen and Pelkonen 1979; Serlin et al 1979; Rendic et al 1979; Pelkonen and Puurunen 1980; Puurunen et al. 1980). These DDIs associated with cimetidine were used as a major advertising point by the competitor introducing another H2 receptor blocker, ranitidine, to the market. Gradually the use of liver preparations for studying potential DDIs of new chemical entities increased, and the drug authorities began to require developers to study potential DDIs before marketing applications (latest versions EMA 2012; FDA 2020). Over the years, some widely published cases, also withdrawals, due to DDIs such as mibefradil, cerivastatin, terfenadine and others highlighted the importance of predictive investigations and the need of validated tools. A tremendous progress of tools has occurred over the last decades and these tools are used increasingly during early drug development and in conjunction with clinical research, especially by pharmaceutical industry and CROs (Fowler et al 2017). Furthermore, various databases and search tools have proliferated to aid research, regulation and clinical work (Grizzle et al 2019).

Now, the question is: are the tools developed for anticipating and predicting CYP-based DDIs effective in the management of DDIs during drug development and in clinical situations? To answer this question, we decided to look at the 10-year period from 2007 to 2016 and assess what kind of information about drug–drug interactions official drug monographs contain. Because two of us (OP, MT) have been involved in surveying inhibitors and inducers (also non-CYP enzymes and the most important transporters) among newly introduced drugs in Finland annually from 2007, we have evaluated this information every year (see the supplementary Tables 1–10) to make a conclusion on whether an individual drug has potential interactions, which clinicians should be aware of when making treatment decisions. At this time, we also searched the literature whether there were any additional information on potential DDIs. However, as the evaluation was performed at the time of the annual publication of the new edition of the physician’s desk reference (Pharmaca Fennica in Finland), it was based mostly on information in the official drug monographs. It should be kept in mind that the supplementary tables concern new drugs authorized in Finland and thus there are some differences as compared to the authorizations in EU (EMA) or USA (FDA) or other authorities. However, practically all the medicines authorized during 2007–2016 in Finland had undergone a centralized process, i.e., approved at the EU level. So overall we believe that the differences between the Finnish and most international pharmaceutical formularies are rather small and that the list of drugs showing potential of CYP inhibition and induction provides an adequate view about the topic. Similar observations have been made by Yu et al (2018, 2019) on FDA-approved drugs in 2013–2017. In the following we present a few observations on the basis of this exercise, summarized in Tables 1 and 2.

Table 1 New drug substances with marketing authorization in Finland for 2007–2016: grouping according to administration, molecular size and some special indications (kinase inhibitors for cancer; anti-HIV drugs)
Full size table
Table 2 New small-molecule drugs with a warning in the monograph that CYP-based interactions are potentially affecting drug treatment and should be taken into consideration
Full size table

Spectrum of new drugs has changed over time

First of all, out of 256 approved drugs, 43% (111) were drugs given parenterally, usually as an intravenous injection or infusion. This group consists of a mixture of products for various indications, but the largest product group is biopharmaceuticals, i.e., biological drugs with special indications such as specific cancers or rheumatoid arthritis. Altogether 17% (43 drugs) of all the approved drugs belonged to this group and they occupy nowadays a major share of new drugs.

57% of the approved drugs (145 drugs) belong to a group of small-molecule pharmaceutics, i.e., “ordinary” drug molecules. Within this group, 24 (17%) belong to novel kinase inhibitor anticancer drugs and 14 (9.7%) to anti-HIV-drugs. Remaining drugs are spread over numerous indications.

CYP substrates, inhibitors and inducers

Out of 145 small-molecule oral drugs, 63 (43%) were substrates of CYP enzymes, predominantly CYP3A4, and 15 (10%) were deemed to be inhibitors of consideration by a regulator and/or developer. Just six (4%) CYP inducers were identified among the new drugs.

Most CYP-associated drugs are either anticancer or HIV drugs

A clear majority among substrates and inhibitors were either anticancer drugs or anti-HIV drugs. Anticancer drugs were mostly kinase inhibitors, which are metabolized principally by CYP3A4. These include bosutinib, dabrafenib, dasatinib, erlotinib, gefitinib, imatinib, labatinib, nilotinib, olaparib, patsopanib, ponatinib, regorafenib, ruxolitinib, seritinib, sorafenib, sunitimib, vandetanib, vemurafenib, and vismodegib. Nine out of 15 CYP inhibitors were kinase inhibitors or anti-HIV drugs. Especially HIV protease inhibitors are variably potent CYP3A4 inhibitors and these include atazanavir, darunavir, fosamprenavir, lopinavir, ritonavir, and saquinavir.

Many non-CYP enzymes and transporters emerge as interaction targets.

As could be seen in the supplementary tables, many transporters have been identified as potential interaction targets for the approved drugs. However, it is often not possible to carefully assess their roles in interactions, because there are only a few validated methodologies available.

CYP3A4 substrates form the major part of the listed drugs

CYPs other than CYP3A4 were only sporadically observed among substrates, inhibitors or inducers. It is remarkable that CYP2D6 was identified as a metabolizing enzyme for only six drugs. CYP1A2 and CYP2C9 were target enzymes for even fewer drugs.

There are only a few inducers

Only six newly approved drugs were CYP inducers over this 10-year period. It seems that the thrust in the development of small molecule drugs has been towards more potent and specific molecules and this has led to a relative decrease of clinical doses, which lead to low hepatic and duodenal concentrations unable to cause a significant CYP induction. Naturally, the induction properties of new molecular entities is studied during the drug development process nowadays and the results guide the process to avoid potential inducers.

There have been no major CYP-DDI surprises leading to drug withdrawals among novel drugs since 2007

It is of interest that after 2007 there are no adversity-based withdrawals that could be clearly and predominantly associated with CYP interactions. Naturally, several non-CYP-associated interactions (e.g., based on P-glycoprotein ABCB1) have been found to be of significance (see some of them in the supplementary tables) and they deserve a proper consideration when assessing the clinical significance of the observed pharmacokinetic consequence. However, as far as we know there have been no withdrawals due to these interactions. It is still necessary to remind of a complex landscape of clinically significant interactions consisting of characteristics of interacting drugs and an individual patient with her/his unique genetic and environmental features.

Conclusion

On the basis of the above analysis, it seems proper to conclude that the predictive tools to investigate CYP-DDIs have been rather efficient in detecting significant interactions and preventing more serious clinical adversities. It should be stressed, however, that any individual pharmacokinetic process cannot be functionally separate or independent from other pharmacokinetic processes (i.e., ADME). Instead, they form a seamless whole, and CYP-associated processes are only a part, although an important one, of the whole process of pharmacokinetics. Consequently, a wider view of the whole process is preferable when judging the possibility and significance of a specific CYP-DDI occurrence.