This study is a post hoc analysis of a prospective, single center, and non-randomized study, which was originally designed to elucidate the mechanism of action of DEB in the setting of ISR. Patients with angina pectoris (both stable and unstable) or silent ischemia, who were scheduled for PCI because of ISR in a BMS or DES were regarded eligible. Exclusion criteria were: acute myocardial infarction, left main disease, ostial ISR (unfit for OCT evaluation), ISR located in a coronary bypass graft, recurrent ISR, presence of renal failure (creatinine ≥200 µmol/L), left ventricular ejection fraction ≤30 % and estimated life expectancy <12-months. The study was approved by the ethics committee of the University Medical Center Utrecht and conducted in compliance with the Declaration of Helsinki. All included patients provided signed informed consent.

All study patients were treated with daily acetylsalicylic acid (80–100 mg) and clopidogrel (75 mg). If not pre-treated, a loading dose of clopidogrel (300–600 mg) was administered before the procedure. Procedural anticoagulation (aimed activated clotting time ≥250 s) was established by intravenous heparin. Administration of glycoprotein IIb/IIIa inhibitors was left at the physicians discretion. After the procedure, acetylsalicylic acid was continued indefinitely and clopidogrel was continued for 1 months.

Treatment of the target lesion was performed by sequential standard balloon predilatation and DEB dilatation. Standard balloons were sized with a 0.9:1 balloon-to-index-stent-diameter ratio, with a length shorter than the intended DEB, and inflated at high pressures (12–19 atm). DEBs were sized with a 1.1:1 balloon-to-index-stent-diameter ratio and inflated at lower pressures (6–10 atm) for 60 s. DEB length was selected to avoid geographic miss (i.e., the DEB should extend ≥5 mm proximal and distal of the predilatation balloon) and undesired DEB overlap in case of multiple DEB use in long lesions. Device characteristics are summarized in Table 1 and elaborated on in the supplementary methods (Online Resource 1). Postdilatation with standard balloons was left at the physicians discretion. Additional stenting was performed in case of stent edge dissection or residual stenosis. Before commencing PCI and directly afterwards, FFR and OCT were performed consecutively. Predilatation with a small diameter (1.5 or 2.0 mm) standard balloon was allowed in case the target lesion could not be crossed with the FFR guidewire or the OCT catheter at baseline. A detailed methodological description of the acquisition and offline analysis of QCA, OCT and FFR data is provided in the supplementary methods (Online Resource 1).

Angiographic follow-up was scheduled per protocol at 6 months, unless indicated earlier on clinical grounds. Clinical follow-up was obtained simultaneously with angiography or by telephone interviews at 6 months. All clinical events were documented after careful examination of relevant hospital files. Antirestenotic efficacy, the outcome of interest, refers to the potency of the DEB to inhibit neointimal growth, not to the prevention of clinical restenosis per se. Main endpoints of this study were angiographic in-segment late luminal loss (LLL) and diameter stenosis, percentage changes in FFR and OCT parameters, and clinical outcomes according to the Academic Research Consortium criteria. Please see the Supplementary Methods (Online Resource 1) for endpoint definitions.

All data were analyzed using IBM SPSS Statistics software version 20 (IBM Corp., Armonk, NY). Continuous variables were presented as mean ± standard deviation or medians (interquartile range), as appropriate. Categorical variables were reported as counts and percentages. Comparison of continuous variables was performed using the Mann–Whitney U test, considering small group sizes. Categorical variables were compared by means of Chi Square or Fisher’s Exact Test. For the most important outcomes, linear regression was used to adjust for differences in relevant baseline and procedural variables between both groups. A two tailed p value ≤ 0.05 was regarded statistically significant.

Angiographic data are presented in Table 4. Baseline angiographic characteristics were similar between the groups. In-segment LLL was significantly smaller for the IN.PACT Falcon than for the DIOR (−0.03 ± 0.43 vs. 0.36 ± 0.48 mm, p = 0.014). The cumulative distribution of in-segment LLL is depicted in Fig. 1 for both DEB. In-segment diameter stenosis at follow-up was lower in IN.PACT Falcon-compared to DIOR-treated patients (30.7 ± 16.2 vs. 41.3 ± 22.6 %, p = 0.083), approaching statistical significance. Binary restenosis (both in-stent and in-segment) rate was 39 % in the DIOR and 17 % in the IN.PACT Falcon group (p = 0.16).

Clinical follow-up at 6-months was available for all patients (Table 4). There were no cases of cardiac death, myocardial infarction or stent thrombosis during follow-up. A strong trend towards a higher TLR rate was observed for the DIOR as compared to the IN.PACT Falcon (35 vs. 8 %, p = 0.057). Of the 7 DIOR and 2 IN.PACT Falcon patients that received TLR, respectively 2 (29 %) and 1 (50 %) presented with DES-ISR at baseline. All lesions treated with TLR were FFR positive.

Fractional flow reserve measurements (Table 4) were performed in 89 % (n = 40), 100 % (n = 45) and 91 % (n = 41) of patients at baseline, postprocedure and follow-up, respectively. Preprocedure FFR was forgone in five patients as the FFR wire could not pass the target lesion, postprocedure FFR was performed in all patients. FFR at follow-up was missing in four patients who denied angiographic follow-up. Baseline and postprocedure FFR distal of the stent were comparable between the groups (p = 0.31 and p = 0.35, respectively), as were in-stent FFR gradients (p = 0.54 and p = 0.40, respectively). At follow-up, FFR distal of the stent was significantly higher in the IN.PACT Falcon group (0.92 ± 0.07 vs. 0.84 ± 0.13, p = 0.029) and in-stent FFR gradient was lower (0.05 ± 0.05 vs. 0.13 ± 0.12, p = 0.002).

Overall, preprocedural, postprocedural, and follow-up OCT (Table 5; Fig. 2) were available in 78 % (n = 35), 96 % (n = 43), and 87 % (n = 39) of patients, respectively, with no differences between the groups. Reasons for missing OCT data were: the inability to cross the target lesion (n = 8) and poor image quality (n = 2), at baseline, technical issues with the acquisition catheter (n = 1) and poor image quality (n = 1) just after the procedure, and the refusal of follow-up angiography (n = 4) and poor image quality (n = 2), at follow-up. Baseline OCT derived lumen, neointima and stent characteristics were well matched between the groups. Postprocedure, mean and minimal stent area were significantly smaller in the DIOR group (both p = 0.02), while residual neointimal burden was comparable among the groups (similar mean and maximum percentage neointimal area stenosis and percentage of stent volume occupied by neointima). At follow-up, mean and maximum neointimal area stenosis as well as neointima occupied stent volume were larger for the DIOR group (all p < 0.05) with concomitant smaller minimal and mean lumen area (both p < 0.01). Stent strut analysis is depicted in Supplementary Table 1 (Online Resource 1). In the IN.PACT Falcon group, a small, but significantly larger portion of overall uncovered stent struts was detected at follow-up (p = 0.001), as practically all struts in the DIOR group were covered.

Table 6 shows the percentage changes in postprocedure and follow-up measurements. A percentage decrease in minimal lumen diameter was observed for the DIOR, while the IN.PACT Falcon showed a relative increase (p = 0.034). Percentage volume changes revealed an overall decrease in lumen and increase in neointima in the DIOR group, as opposed to an increase in lumen and decrease in neointima in the IN.PACT Falcon group (Fig. 3). A significant difference was observed among the DEB in the percentage change of in-stent FFR gradient between postprocedure and follow-up, demonstrating an in-stent gradient increase for the DIOR and a decrease for the IN.PACT Falcon (p = 0.003). The difference in neointimal growth and lumen volume change on OCT remained statically significant after adjusting for hypercholesterolemia, postdilatation, and postprocedure minimal lumen area on OCT. Comparable differences were observed when focusing only on BMS-ISR (see Supplementary Table 2, Online Resource 1).

Antirestenotic efficacy was higher for the IN.PACT Falcon compared to the DIOR, which resulted in a benefit in clinical outcome. The present results for the IN.PACT Falcon are in line with those previously reported for this particular DEB. A small prospective study on BMS-ISR observed an in-segment LLL of 0.02 ± 050 mm and TLR rate of 4.3 % at 6 months [20]. In this regard, the IN.PACT Falcon compares favorably to other benchmark DEB in the setting of ISR [10, 21]. The findings for the DIOR are on odds with prior data. In the Valentines-I trial, a prospective study on the efficacy of the DIOR in both DES-ISR and BMS-ISR, an encouraging 7.4 % TLR rate was found after a mean follow-up of 7.5 months [22]. The lack of mandatory angiographic and FFR follow-up may explain the discrepancy with the present study findings.

Notably, the larger proportion of drug-eluting index stents in the DIOR group had raised some concern for bias, as recurrent restenosis is more common in DES-ISR compared to BMS-ISR [2, 22]. Therefore, a subanalysis in BMS-ISR patients was performed, which showed results identical to the overall population (supplementary Table 2, Online Resource 1). Other factors that have been associated with recurrent restenosis, i.e., lesion length and ISR pattern [23], were well balanced between the groups. Study design did not account for unmeasured confounding factors, such as the pathophysiology and tissue characteristics of ISR. Both features are closely related to the type of index stent, however [24].

The present study confirms, for the first time using mandatory angiographic follow-up, that no class effect can be assumed for DEB in the clinical setting. An earlier report from the SCAAR (Swedish Coronary and Angioplasty) registry has shown important differences in restenosis between two commonly used DEB [2]. Of note, the DEB compared in SCAAR were very dissimilar with respect to the employed modifications to improve drug release, as the DEB demonstrating the worse outcome forgoes the use of an excipient. The pivotal role of excipients in DEB was demonstrated before in preclinical studies [13, 14] and is reflected in the ambiguous results of clinical trials in small vessel disease [16, 17].

Despite both DEB in our study are equipped with an excipient-based coating formulation, large differences in antirestenotic efficacy are observed. This is best expressed by the percentage changes in neointimal volume during follow-up, revealing a 30 % increase for the DIOR and a 16 % decrease for the IN.PACT Falcon (Fig. 2). Where the IN.PACT Falcon not only inhibits neointimal formation, but also induces positive remodeling, the DIOR fails to suppress neointimal growth to a sufficient degree. Positive remodeling after DEB treatment may result from neointimal smooth muscle cell loss due to paclitaxel-induced apoptosis or necrosis, two phenomena that require high local tissue concentrations of drug [25]. Otherwise, cicatricial shrinkage of neointima through healing of the barotrauma-induced dissections may be involved [19]. Regardless of the underlying mechanism, the net effect of these processes only becomes apparent as positive remodeling in the presence of sufficient neointimal suppression.

The observed disparity in antirestenotic efficacy may be attributed to the difference in excipients, since all other DEB characteristics were constant for both devices. Animal studies have suggested before that the use of an excipient in a DEB coating is no guarantee for effective drug transfer to the vessel wall [14]. However, early paclitaxel tissue concentrations measured during the preclinical testing of the DIOR were similar to those reported for other established DEB [26, 27].

According to more recent data, it may not be the acute transfer of drug, but the durability of paclitaxel activity that falls short in the DIOR. Using histological vascular healing parameters (inflammation-, fibrin- and smooth muscle cell loss score) as a proxy, the drug effect in DIOR treated porcine arteries was observed to disappear rapidly as it was already absent at one month follow-up [28]. This is unsettlingly fast, as the vascular healing signs associated with DEB well-known for their effectiveness remain even detectable up to several months [29]. Thus, although effective drug uptake does occur in DIOR angioplasty, its effect appears short-lived. The cause of this unfavorable pharmacokinetic profile of the DIOR remains speculative, but may reside in an undesired drug-excipient interaction that may lead to: (1) a decreased binding of paclitaxel to non-specific binding sites accelerating its tissue clearance, (2) a chemical configuration of paclitaxel that interferes with its metabolic activity, or (3) less homogeneous release of paclitaxel resulting in unfavorably large spatial differences in tissue drug concentration. Interestingly, the above data supports the longstanding presence of active drug as the mechanism responsible for durable neointimal inhibition by DEB, rather than the prevention of early growth initiating events [27].

This study is the first in-human comparative assessment of DEB using highly sensitive techniques to quantify morphological and functional changes at the target lesion site. The unique investigational approach ensued, allowed for a precise and reliable assessment of antirestenotic efficacy.

While angiography is commonly used to assess neointimal inhibition after DEB treatment, it remains lumenography: an imaging technique providing data on lumen diameter change at follow-up without differentiating between the underlying mechanisms (neointimal growth or stent recoil). In contrast, morphological assessment by OCT provides separate measurements on lumen, neointimal and stent dimensions. This enables the precise calculation of neointimal dimension changes between specified time points, irrespective of changes in stent volume. Moreover, the OCT derived neointimal volume change at follow-up represents neointimal inhibition along the entire lesion length, whereas its angiographic counterpart (LLL) may only reflect local neointimal suppression. Neointimal volume change on OCT may thus provide a more honest assessment of antirestenotic potency, although focal lumen changes may be ultimately responsible for recurrent restenosis.

Finally, structural functional assessment at follow-up by means of FFR allowed for objective evaluation of the impact of morphological changes on vessel patency. This approach is highly recommendable, since a poor correlation exists between angiographic assessment of moderate or diffuse type ISR and the hemodynamic significance of lesions assessed by FFR [30]. FFR confirmed the better antirestenotic efficacy for the IN.PACT Falcon observed on morphological data.