Imaging microvascular changes in nonocular oncological clinical applications by optical coherence tomography angiography: a literature review

A pioneering effort in assessing microvasculature by means of OCTA in clinical oncology was the work of Tsai et al.¹⁵ They applied a modality to image subsurface vascular patterns in a patient with nondysplastic Barrett's esophagus (NDBE) and demonstrated that in this way, the diagnostic capability of endoscopic OCT was enhanced. Lee et al.¹⁶ continued their work and collected 97 datasets from 52 patients with NDBE and low-grade/high-grade dysplasia (LGD/HGD). Their goal was to differentiate NDBE patients from LGD/HGD patients; however, due to insufficient image quality, 43 datasets (44%) in 20 patients were not used for analysis; OCTA images were also not generated in real time due to the high computational burden. The findings of the study revealed distinct differences in microvascular OCTA features of abnormal vessel branching and heterogeneous vessel size between the NDBE and LGD/HGD groups, as shown in Figure 2. Further research with a larger patient population is required to validate these findings and establish the clinical utility of endoscopic OCTA in routine practice.

FIGURE 2.

In the study by Maslennikova et al.¹⁷, clinicians aimed to investigate the use of OCTA for imaging microvascular changes in the oral mucosa of cancer patients undergoing radiotherapy (RT). Authors conducted a longitudinal study involving 25 patients with oropharyngeal and nasopharyngeal cancer undergoing RT. OCTA was employed to visualize and analyze the microvascular network within the oral mucosa over time, and imaging was performed before the treatment and at regular intervals during and after RT. OCTA images were generated in real time by the acquisition system shown in Figure 3. The findings of the study demonstrated significant alterations in the microvascular morphology and density in the irradiated oral mucosa over the course of RT, with the microvascular network showing an increase in the vascular density and total length of capillary-like vessels compared to the baseline measurements. These changes were found to be more prominent when grade two and three mucositis developed. The study demonstrated the potential of OCTA as a valuable tool for longitudinal monitoring of microvascular changes in radiation-induced oral mucosal damage. However, it is important to note that this study has several limitations, as the sample size was relatively small, and the results may not be generalizable to the broader population; additionally, the prognostic significance of the observed microvascular changes needs further investigation.

FIGURE 3.

De Carvalho et al.¹⁸ published a case report in which they showed an increased vasculature in the melanoma region compared to the nevus. Following this, Themstrup et al.¹⁹ conducted a study to distinguish subtypes within the keratinocyte skin cancer spectrum enrolling 18 patients with actinic keratosis (AK), 12 patients with Bowen's disease (BD) and 24 patients with squamous cell carcinoma (SCC). In this exploratory clinical study, they identified two vascular features that showed significant differences across the lesion types. One of these vascular features, referred to as “blobs”, i.e., small, isolated points with a simple round appearance, was more frequently present in BD cases but either absent or only slightly present in AK and SCC lesions. The other feature, called “curves”, i.e., narrow, curved, continuous structures of varying length, was predominantly present in AK lesions. These findings are illustrated in Figure 4.

FIGURE 4.

In a subsequent study²⁰, the same group continued with the differentiation of common basal cell carcinoma (BCC) subtypes by scanning 81 patients with 98 BCC lesions, of which 27 were superficial BCC (sBCC), 55 were nodular BCC (nBCC) and 16 were infiltrative BCC (iBCC). In this study, they found various structural and microvascular features that would aid in identifying nBCC, iBCC and sBCC subtypes. For example, it was shown that the presence of so-called “serpiginous” vessels, i.e., wavy structures of varying length, indicated an increased risk of nBCC and a reduced risk of sBCC.

Meiburger et al.²¹ applied OCTA to a patient with nBCC and six patients with sBCC and developed an algorithm for automatically determining skin lesion area using vascular density. While authors were hopeful in their conclusion that proposed method could facilitate diagnosis and treatment of BCC, no further study was published.

Gubarkova et al.²² examined 27 patients with BCC who received photodynamic therapy (PDT). They utilized OCTA imaging before and immediately after PDT and during follow-up visits to monitor vascular changes. Analysis of the OCTA images allowed for quantification of parameters such as blood vessel density and uniformity, aiding in distinguishing among BCC subtypes. The study demonstrated that OCTA offers real-time information on vascular changes in response to PDT. The researchers observed a decrease in blood vessel density at 24 hours after PDT, with OCTA images having 97% predictive value for differentiation between complete and partial responders.

De Carvalho et al.²³ conducted a systematic analysis of melanoma lesions in 127 patients and found a significant link between specific microvascular features and Breslow's thickness. In a more recent study, Welzel et al.²⁴ assessed 159 melanomas from 156 consecutive patients and found that irregular vascular shapes, including blobs, curves and serpiginious vessels, were more common in high-risk and metastatic melanomas than in low-risk lesions. Most recently, the same group²⁵ prospectively examined a total of 167 nevi, including dysplastic ones, in 130 participants and compared these microvascular features to those found earlier in 159 melanomas.²⁴ They found that increased blood vessel density and diameter and irregular tissue architecture were associated with melanomas, while nevi showed more regular structures and lower blood vessel density and diameter, indicating their benign nature (Figure 5). Researchers also found excellent predictive diagnostic value of microvascular features (e.g., blobs, serpiginious vessels) for nevi (88.2% to 91%) and melanoma (95.5% to 96.8%) and concluded that OCTA “may be a valuable addition to the current clinical-dermoscopic gold standard”.²⁵

FIGURE 5.

One of the obvious limitations of OCTA in nonocular clinical oncology settings is its restricted penetration depth. OCTA relies on detecting motion contrast generated by moving RBCs, which limits its applicability to superficial structures. Tumors and lesions in deeper anatomical locations, such as within organs or soft tissues, may not be adequately visualized using OCTA due to limited tissue penetration. This constraint hampers its potential for comprehensive evaluation and monitoring of oncological conditions.

However, this limitation can be overcome by using endoscopic techniques bringing the instrument closer to the tissue of interest. As demonstrated in the GI tract studies by Tsai et al.¹⁵ and Lee et al.¹⁶, OCTA can be used endoscopically. Namely, common OCT has been developed as endoscopic probes of different types and used to obtain microscopy images of entire luminal organs, solid tumors, or vessels.²⁶ Since OCTA is an extension of OCT, the same already developed technology can be used to bring the system closer to the tissue of interest.

Another possibility to increase the OCTA penetration depth is to use OCTA systems with longer wavelengths. In the studies presented in this article, the OCTA systems utilized 1.3 μm wavelengths, which is a typical wavelength also used for skin imaging; in ophthalmology, a shorter wavelength of 0.8 μm is typically used, resulting in an approximately 60% lower penetration depth. In a recent publication by Nishizawa and Yamanaka²⁷, it was shown that by using a 1.7 μm wavelength, the penetration depth increases by approximately 40% compared to a 1.3 μm wavelength. Therefore, by developing OCTA systems with even longer wavelengths, larger penetration depths could be obtained.

OCTA provides detailed structural information about blood vessels but lacks the ability to differentiate between different vessel types. In the field of oncology, the distinction between arterial and venous vasculatures is crucial, as tumor angiogenesis is primarily associated with the growth of new abnormal blood vessels. Accurate differentiation between these types of vessels aids in assessing tumor progression and treatment response. Unfortunately, OCTA's current capabilities fail to provide this level of vessel characterization, limiting its effectiveness in nonocular oncological settings.

In ophthalmology, recent articles report the possibility of differentiating between arteries and veins utilizing various OCTA image parameters, including vascular diameters and shape and perfusion intensity density.²⁸ However, the current methods for artery-vein classification in OCTA employ complex algorithms, thereby making it difficult for clinical applications. To alleviate this hindrance, deep learning algorithms were developed to reduce the complexity and automate artery-vein classification.²⁹ Similar algorithms should also be developed for other OCTA modalities.

Movement, including patient motion during OCTA acquisition, can introduce motion artifacts, leading to image distortions and reduced image quality. Unlike ophthalmology, where patients can fixate on a target, patients in nonocular oncology settings often have limited control over motion, making motion artifacts more challenging to mitigate. This limitation can compromise the accuracy and reliability of OCTA in nonocular clinical oncology, demanding the need for advanced postprocessing algorithms to improve image quality.

Since motion artifacts are well-known sources of artifacts in OCT imaging, they have been extensively researched. One possibility is to detect and compensate for the axial motion artifacts pixelwise by comparing the topology of different layers in tissue, and the motion artifacts are then compensated by shifting the pixel numbers with the value detected.²⁹ Another possibility is to remove the affected scans in the software and to use only the nonaffected scans for vasculature image reconstruction.³⁰ However, this approach may increase the duration of imaging sessions; therefore, it would be better to use an approach without the need for rescanning. As a solution, it was demonstrated that the motion contribution to the OCT signal can be reasonably estimated by considering statistics of the measured flow signal across all voxels.³⁰ By implementing motion artifact compensation strategies, the translation of OCTA to clinical workflow would become more feasible.

The lack of standardized protocols and interpretation guidelines is a significant limitation of OCTA in nonocular clinical oncology. Unlike ophthalmology, where standardized imaging protocols and interpretation criteria exist, the application of OCTA in oncology lacks such standardization. As a result, different centers may use varying acquisition settings, image processing algorithms, or interpretation approaches, leading to inconsistent and noncomparable results. Establishing standardized protocols and guidelines specific to nonocular oncology would enhance the accuracy and reproducibility of OCTA findings.

While OCTA has shown great promise in ophthalmology, its translation to the nonocular clinical oncology setting faces limitations. In particular, the lack of standardized protocols and interpretation guidelines poses a significant challenge. Addressing these limitations through advancements in technology, algorithm development, and a larger number of clinical sites initiating clinical trials is essential for realizing the full potential of OCTA in nonocular clinical oncology.