The primary findings of this study underscore the centrality of anatomical parameters derived from angiographic three-dimensional QCA in predicting the occurrence of adverse coronary events at a 5-year mark. This research bridges the gap between angiography and intracoronary physiology, providing new insights into the interplay of anatomical and functional metrics in determining the risk of future MI. In particular, we observed significant differences in the MLR, MLA, ΔvFFR, and vFFR values between mild coronary lesions that will be the site or not of a subsequent acute coronary event. Among these parameters, MLR emerged as the strongest predictor of MI at 5 years. Lesions characterized by lower MLR values, hence presenting with a sharper luminal narrowing, exhibited a substantially increased risk of destabilization.

Previous studies suggested that the severity of luminal narrowing at the site of a lesion is critical in the pathogenesis of plaque progression and destabilization and is predictive of adverse events at the follow-up [13, 14]. Area stenosis and MLA—defined either angiographically or by intravascular imaging—emerged as established marks of CAD severity [15, 16]. More recently, plaque components and perivascular tissue elements were clinically linked with lesion vulnerability and adverse events [14, 17, 18].

In the present study, anatomical features of lesions exiting clinical quiescence and provoking overt myocardial events were compared to those of quiescent ones from the same patients, thereby providing a patient-specific control for systemic biological confounders. As a result, the ratio between the cross-sectional areas at the lesion’s proximal edge and at the lesion’s narrowest part, captured by the MLR, provided a robust prediction for future adverse coronary events, outperforming traditional anatomical descriptors of lesion severity. In fact, by mathematically adjusting the narrowest lesion cross-sectional area (i.e., the MLA) with the cross-sectional area of the lesion proximal edge, MLR encompasses the antegrade disease progression along the vessel. In this way, MLR accounts for the narrowing of the diseased vessel segment. Considering previously published data on 5-year predictivity for MI in the same patient population, MLR showed to overcome percental area stenosis (%AS) in performance (%AS AUC = 0.65 (95% CI 0.57–0.73) vs. MLR AUC = 0.75 (95% CI 0.68–0.82)) [6].

Besides that, rather than the disease’s progression along the vessel’s axis, it was the radial inward protrusion of the vessel wall and the subsequent luminal narrowing that correlated with the studied clinical outcomes. In fact, anatomical descriptors of lesion length, including PLSL, LSL, and LLR, did not differ among the FCL and NCL groups and presented no predictive capacity for future MI, while MLR and MLA emerged as significant predictors. The lack of significant variance in PLSL between the FCL and NCL groups particularly highlights the MLR’s independence from lesion length information when assessing myocardial infarction risk. These insights suggest that focusing on the severity of luminal narrowing, rather than the length of the lesion, might be more critical in predicting myocardial events.

Notably, the methods of the present investigation and the MLR computation could be translated to any invasive and non-invasive imaging modality able to produce a 3D vessel reconstruction (including computer tomography angiography, optical coherence tomography, and intravascular ultrasound). The here proposed angiography-based approach might still represent the optimal compromise between the accuracy of vessel reconstructions (lower with non-invasive imaging modalities) and costs (higher in invasive intra-vascular imaging) [19].

Intracoronary pressure measurement has evolved from a static distal measure (FFR) to a dynamic evaluation along the length of the vessel, thanks to pullback pressure measurements [20]. The ability to functionally identify focal or diffuse CAD phenotypes by analyzing the loss of pressure along the wired vessel has given rise to the concept of local pressure drop patterns. The focal CAD pattern was associated with higher plaque burden, lipid-rich composition, and vulnerability traits [21] and correlated with better angina relief post-percutaneous treatments when compared with the diffuse CAD phenotype [22].

From a functional standpoint, in our study, a pronounced lesion narrowing (identified by a low MLR) was associated with a focal pressure loss pattern along the lesion (evidenced by a high ∆vFFR) as well as with a larger pressure decay along the vessel (denoted by a low vFFR). Moreover, based on the Bernoulli theorem, the lumen narrowing in the converging flow segment of the lesion, quantified by MLR, imparts blood flow energy transformations [23]. The amount of such blood flow energy transformations established within the lesion is expected to impart/exacerbate flow disturbances. These disturbances can be distilled into wall shear stress (WSS) profiles acting at the lesion level, which have been recently identified as strong predictors of lesions culprits of future MI in the here investigated dataset [6, 23].

These considerations suggested that MLR, while deriving from the geometric features of the stenosis, inherently carries a functional significance, which could in turn enhance its clinical predictive potential. Accordingly, in our study, MLR demonstrated superior predictive capacity for the clinical endpoint not only to other anatomical parameters but also to the evaluated functional lesion severity descriptors, such as ΔvFFR and vFFR.

This study is not without limitations. It is retrospective in nature, and therefore subject to potential selection bias. The study population is also specific to patients who presented with an acute MI and had prior coronary angiography, which may limit the generalizability of the findings. Furthermore, the potential longitudinal variation and time-dependent effects of cardiovascular risk factors may have influenced our results. However, the inclusion of NCL as an internal control within our study design likely mitigated the impact of these potential confounders. Finally, while our analyses were blinded to the FCL/NCL classification, future prospective studies that incorporate these parameters into the design would further solidify our understanding of these complex relationships.