**5. Drug–Drug Interactions**

Under consideration of the high number of frequently prescribed drugs that were described to be substrates (Table 3), inhibitors (Table 4), or inducers (Table 2) of human CYP1A1/A2, several unwanted drug–drug interactions can be assumed in the case of combined administration.

### *5.1. Inhibition Studies*

In this regard, the most pronounced interactions have been described for the combination of CYP1A substrates with potent inhibitors including ciprofloxacin, fluvoxamine, ethinyl estradiol, and rofecoxib. Their combination with established CYP1A substrates resulted in clinically relevant interactions increasing the systemic drug exposure of caffeine, clozapine, mirtazapine, olanzapine and theophylline by 1.5 to 3-fold [106,131,142–144]. For agomelatine, ramelteon, tracrine, and tizanidine much more dramatic increases of serum area under the concentration-time curve (AUC) by 10–190-fold have been observed [118,119,145,146], which is expected to cause drug-related side effects and even toxicity. For example, the elevation of plasma levels of clozapine by ciprofloxacine resulted in rhabdomyolysis, delirium, and death during combination in psychotic patients [147,148].

The reasons for these dramatic interactions might be due to extensive metabolism by CYP1A enzymes and/or a high volume of distribution of the victim drug (e.g., 168 and 349 l for tizanidine and tacrine). In order to estimate the in vivo potential of a certain CYP1A inhibitor (Table 4) of an in vitro function, to cause clinically relevant interactions, focusing on the observed inhibitory potential (Ki, IC50 value) alone is not sufficient, but additional pharmacokinetic aspect of the perpetrator compounds must be considered as well. For sufficient inhibitory potential in vivo, a perpetrator drug needs to

generate free unbound concentrations (fraction unbound, fu) around or above the observed Ki/IC50 value and needs to be present in the systemic circulation for several hours to cause substantial metabolic inhibition as determined by an elimination half-life of several hours. Consequently, fluvoxamine and ciprofloxacin that are characterized by rather low-to-medium protein binding (fu, 0.23 for fluvoxamine and fu, 0.8 for ciprofloxacin), but high serum levels as caused by their comparatively high administered doses (50–100 mg for fluvoxamine, 100–750 mg for ciprofloxacin) and medium to long terminal half-lives (4–7 h for ciprofloxacin, 17–22 h for fluvoxamine), cause that both drugs are strong inhibitors of CYP1A2 in vivo, and cause many clinically relevant drug–drug interactions.

This scenario is not true for other drugs mentioned in Table 4. For example, although the NSAIDs celecoxib and tolfenamic acid demonstrated a considerable inhibition of CYP1A2 in human liver microsomes (HLM) with a Ki values of 25 μM and 1.4 μM [127], they did not show clinically relevant interactions, most likely due to their high protein binding of ~98% and 99.7%, respectively. As a conclusion, drugs undergoing substantial CYP1A1/2 metabolism should be combined with caution together with the perpetrator drugs mentioned in Table 5. If possible, dose escalation combined with therapeutic drug monitoring should be used for CYP1A2 drugs with a narrow therapeutic index such as theophylline, clozapine or tizanidine. Whether the mentioned in vivo inhibitors of CYP1A2 may also cause clinically relevant interactions with CYP1A1 substrates remains uncertain and requires further studies.


**Table 5.** Overview of clinically relevant interaction as caused by inhibition of CYP1A1/1A2 enzymes.


↑, increase; ↓, decrease; AUC, area under the concentration-time curve; BID, twice daily; CL, clearance; Cmax, maximum serum concentration; Css, trough serum concentrations at steady-state; d, days; MD, multiple doses; PK, pharmacokinetic; SID, once daily; SD, single dose; t1/2, elimination half-life.

### *5.2. Induction Studies*

On the other side, carbamazepine, lansoprazole, omeprazole, phenobarbital, primaquine, and rosiglitazone were reported to be potent inducers of CYP1A1/1A2 by binding to AhR or CAR receptor as briefly described above [44,46–49] (Table 2), while the effects of prototypical PXR activators such as rifampicin, ritonavir and St. John's wort are rather negligible [47,58–60,156]. Of these drugs, omeprazole was one of the most potent and most extensively investigated inducer in vitro and in vivo, resulting in several-fold induction of the gene expression, protein abundance and metabolic function of CYP1A1/2. However, significant effects on the pharmacokinetics and efficacy of CYP1A substrates have not been observed yet. Well-established substrates, including caffeine, phenacetin, theophylline, or propranolol did not show any changes in their pharmacokinetics in the presence of omeprazole [157–160]. Thus, one might conclude that the interaction potential of omeprazole and other proton pump inhibitors for clinically relevant DDIs might be very limited although there are also data from a case report indicating slight increase in CYP1A2 metabolism [161]. An explanation could be found in the relatively low peak concentrations of omeprazole (0.7–4.6 μM) in the systemic circulation compared to the inductive in vitro concentrations (25–50 μM) and its short half-life of 0.5–1 h (Regardh et al. 1990). In contrast to this, treatment with carbamazepine considerably induced clozapine metabolism, leading to significantly lower serum level in schizophrenic patients [149]. Carbamazepine was furthermore shown to induce hepatic caffeine metabolism as well as the systemic clearance of olanzapine and mirtazapine in a significant manner [55,162,163]. Thus, it can be stated that CYP1A2 substrates should not be combined with carbamazepine or dose-adjustment should be taken into account.

However, estimations on potential drug interactions using in vitro data on induction properties alone can be misleading. An example for that phenomenon is ritonavir, a HIV protease inhibitor. Although it showed no (or only weak) induction of CYP1A2 mRNA and activity in human hepatocytes [58,61], the pharmacokinetics of caffeine and olanzapine was significantly affected, i.e., AUC was reduced by 75% and 53% [164,165]. To overcome decreased drug efficacy due to the considerable changes in the pharmacokinetics, Jacobs et al. (2014) proposed that doubling the dose of olanzapine as a successful strategy in the case of co-medication with ritonavir [166]. The same disconnection between in vitro and in vivo effects could be observed for rifampicin, which has not been shown to be an AhR ligand and demonstrated also only a weak induction of CYP1A2 expression and metabolic function in human hepatocytes [46,47,59,61]. Accordingly, rifampicin premedication for 5–15 days reduced serum AUC of caffeine and tizanidine by 50–60% [64,164]. The reasons for this surprising finding might rely on nuclear receptor cross-talk or insensitivity of the respective in vitro on nuclear receptor activation [167]. Table 6 provides an overview about clinically relevant interactions of CYP1A substrates caused by enzyme induction.


**Table 6.** Overview of clinically relevant interaction as caused by induction of CYP1A1/1A2 enzymes.

↑, increase; ↓, decrease; AUC, area under the concentration-time curve; BID, twice daily; CL, clearance; Cmax, maximum serum concentration; Css, trough serum concentrations at steady-state; d, days; MD, multiple doses; MR, metabolic ratio; PK, pharmacokinetic; SID, once daily; SD, single dose; t1/2, elimination half-life.

Although it was shown in vitro experiments that CYP1A1 can be inhibited and induced by several compounds (Tables 2 and 4), there are to the best of our knowledge no clinical drug–drug interactions that can be attributed by specific CYP1A1 inhibition or induction. However, considering the overlap in substrate recognition, inhibitors, and inducers one might speculate similar interactions as described for CYP1A2 substrates (Tables 5 and 6). Accordingly, relevant DDIs have been estimated for CYP1A1 [179]. Nevertheless, given the low expression levels of CYP1A1 in human intestine and liver (if even), the extent of these interactions is expected to be much lower than those caused by inhibition or induction of hepatic CYP1A2.

#### *5.3. Impact of Smoking and Diet*

Finally, smoking can have a profound effect on the pharmacokinetics and efficacy of several CYP1A1/2 substrates, which is comparable to potent inducing drugs, as summarized in Table 6. In all summarized examples, the systemic drug exposure of CYP1A substrates was significantly decreased in smokers compared to nonsmokers by 30–70% (Figure 2). Thus, smokers require higher doses than nonsmokers. A questionable benefit might be that smokers show also less adverse drug reactions than nonsmokers [177,180]. However, this is so far only well established in neuropsychopharmacology, i.e., treatment with antipsychotics and antidepressants. Here, individual dose adjustment is routinely performed in dependence on therapeutic drug monitoring [181].

ƚŝŵĞ;ĚͿ

**Figure 2.** Schematic overview of potential interaction scenarios of tobacco smoking. In scenario 1, a nonsmoker reaches the steady state conditions of a certain CYP1A1/2 substrate after 5–6 half-lives using standard doses. After start smoking, CYP1A enzymes are significantly induced in intestine and liver resulting in increased drug clearance and decreasing plasma levels of the respective drug. In scenario 2, a smoker, who has already substantially higher expression and metabolic activity of CYP1A1/2, requires significantly higher doses to reach the therapeutic range. After stopping smoking, CYP1A1/2 will gradually return to the native expression levels, while the daily drug doses are not adjusted, which results in markedly increased and potentially toxic plasma concentrations.

It was shown that tobacco consumption induces CYP1A2 activity in a dose-dependent manner; smoking of daily 1–5, 6–10 and >10 cigarettes increases CYP1A2 activity 1.2-, 1.5- and 1.7-fold [182]. The maximum induction effect is already reached after smoking about 10 cigarettes daily, which abates after about three days of stopping smoking [182,183]. In particular, the latter effect may cause safety issues in the case of treatment with highly CYP1A-metabolized drugs with serious side effects, such as clozapine, olanzapine, tacrine, theophylline, or tizanidine. In this case, systemic drug exposure will substantially increase due to decreasing metabolic capacity, but unchanged high doses associated with an augmented risk for side effects, or even drug-related toxicity (Figure 2). Associated with this, cases of agranulocytosis and seizures have been reported for clozapine [181,184]. Because nicotine alone does not possess any inductive effects on CYP1A, the same risk is true in case of using e-cigarettes and other ways of nicotine substitution [185]. This should be considered by adjusting the appropriate dose, especially in case of changes in smoking habit (Figure 2).

Although a chargrilled meat diet was shown to significantly induce intestinal CYP1A1 protein as well as the metabolic activity of hepatic CYP1A2, as concluded from the caffeine breath test [33], altered pharmacokinetics of tacrine and caffeine could not be observed in a respective clinical study [186]. Some in vivo findings sugges<sup>t</sup> also a potential in vivo inducing effects of broccoli [187] and another brassica vegetable, kale [188], on CYP1A2 mediated metabolism of caffeine. The brassica vegetable CYP1A2 induction is most probably mediated by 3,3-diindolylmethane (DIM), a condensation product of indole-3-carbinol being a metabolite of the indole glucosinolate glucobrassicin. DIM has been shown to induce CYP1A2 in cultured human liver slices [189]. However, there is a lack of information about brassica vegetables interaction with clinically relevant drugs.

#### **6. Summary and Conclusions**

CYP1A1 and CYP1A2 are expressed in human intestine and liver. However, their intersubject expression and function is highly variable as most likely caused by genetic, epigenetic, environmental factors (smoking, diet) and diseases. Considering the high number of drugs that have been identified as substrates, inhibitors, or inducers of CYP1A enzymes, many clinically relevant interactions have been reported and can be expected for other substrates. Thus, respective combinations should be avoided or appropriate dose adjustment is recommended in case of victim drugs with a narrow therapeutic index. In general, there is a substantial lack of data regarding CYP1A1 and its distinct role in the pharmacokinetics of drugs. However, from today's perspective, its allover contribution to serious drug–drug interactions seems to be limited considering its low expression levels and the potential functional compensation by CYP1A2. On the other hand, CYP1A2 has to be considered as one of the big five hepatic drug metabolizing enzymes (along with CYP3A4, CYP2C9, CYP2C19, CYP2D6, CYP2E1), which is of high clinical relevance in terms of inter-subject variability of drug e fficacy and safety, as well as drug–drug interactions.

**Author Contributions:** Conceptualization, investigation, writing—original draft preparation, and writing—review and editing, F.K., C.W., M.D., and S.O. All authors have read and agreed to the published version of the manuscript.

**Funding:** The research was funded by the German federal ministry of economic a ffairs and energy (ZIM, project number: 16KN077223) and by an institutional grants of the Institute of Pharmacology and Toxicology, Rostock University Medical Center, Rostock, Germany.

**Conflicts of Interest:** The authors declare no conflict of interest.
