In recent years, the evaluation of the posterior vitreous cortex status has become of great interest, while it can be decisive in assessing the best treatment modality for various retinal conditions [3, 6,7,8,9]. Only a few studies that have compared the preoperative reliability of diagnostic methods with intraoperative findings [9,10,11, 14, 16]. In this study, we have quantified the ability of BM, US, and SD-OCT to diagnose complete PVD compared to triamcinolone assisted intraoperative assessment in patients with different vitreoretinal interface disorders. To the best of our knowledge, this is the first prospective study focusing on the detection of complete PVD with all three methods compared to with intraoperative PPV in a relatively large sample of patients with ERM and FTMH.

The main finding of our study is that the three diagnostic procedures were unable to provide accurate information on PVD status compared to the intraoperative PPV findings. This result is consistent with the conclusions of earlier research [9, 11, 14, 16]. In this study, US was the most accurate non-invasive technique for assesing PVD before PPV, which confirms findings previously reported by Kičova et al. [9]. Although triamcinolone-assisted PPV is considered a gold standard for posterior vitreous visualization [16], it is also associated with a lower rate of reoperations [36]. Furthermore, considering the low sensitivity of preoperative diagnostic methods, the use of triamcinolone staining during PPV might be imperative to achieve adequate vitrectomy, resulting in the best possible postoperative anatomical and functional results [16].

The results of the current study indicate that complete PVD is less common, with only a quarter of all patients presenting this finding. Previous studies that analysed the reliability of different diagnostic modalities showed significantly different frequencies of PVD detected during PPV, ranging from 16 to 60% [5, 9, 11, 14]. It is important to note that these studies used different definitions of PVD staging, included different numbers of patients with various indications and preoperative characteristics. However, in patients with FTMH complete PVD is found in approximately 21% of patients [37], which is consistent with 19.7% presented in this study. On the other hand, in cases with ERM, the reported prevalence of intraoperative PVD is between 77.3% and 79.9% [38, 39], which is significantly higher than the 31.6% presented in this study. A possible explanation might be based on the concept of a complete and incomplete PVD [1], i.e. the criteria that were used in previous research. Histopathological studies confirmed that remnants of the vitreous cortex membrane frequently remain attached to the fovea after apparent complete PVD [40]. Unlike previous studies, all patients included in this study underwent complete vitrectomy with triamcinolone staining and ILM peeling. In this manner, we were able to detect and clean all vitreous remnants and reliably determine the state of PVD in all of our cases. Another explanation can be found in the selection of patients included in previous studies, while differences between studies in patient age, race, sex, or prevalence of systemic (diabetes) and ocular (myopia, pseudophakia) factors can certainly have an impact on PVD occurrence [41].

In this study BM has presented a sensitivity of 48.4% to determine the status of PVD by presence of the Weiss ring. The results of previous studies have shown significantly higher sensitivity of 76% [9] and 90% [14], in identifying complete PVD. Although BM can provide a dynamic, wide-angle observation of the posterior pole [14], it is also considered to be an investigator dependent method that is largely dependent on patient cooperation and transparency of the ocular media [9, 15]. In their study protocol, Stavrakas et al. [14] classified all dubious cases as non PVD during the BM examination. However, this protocol was not used in our study, and we have presented a high rate of false positive results and, consequently, a poor sensitivity. On the other hand, the presented specificity of 81.5% could be considered approximately equal to the results of previous research. This could imply, that in cases where the Weiss ring is not clearly visible during BM, it is more likely to be a partial PVD rather than a complete PVD.

Ultrasonography previously reported a sensitivity of 83% in the detection of complete PVD [8], which is better than 61.3% presented in the present study. This difference may be due to several factors. Interpretation of US findings could be subjective and dependent on the examiner’s ability to detect dynamic vitreous movements [9, 12, 15]. Furthermore, due to its relatively low resolution, even with higher gain and direct ocular contact, US has limitations regarding the detection of flat PVD [12]. This probably resulted in a high number of false negative findings and poor sensitivity in the present study, although we used dynamic ultrasonography in all of our cases. Nevertheless, the reported specificity of 90.2% could be considered within the scale of previous reports [9].

The application of OCT is useful for identifying various preretinal, retinal, subretinal, and choroidal changes. High-quality OCT images can serve as an excellent screening tool but also help to monitor response to different therapeutic modalities [17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32]. In the analysis of complete PVD, time domain OCT provided a correct assignment in only 12.5% of the cases and reported a high percentage (73.3%) of inadequate evaluations [9]. On the other hand, Hwang et al. used 6 mm SD-OCT images with visualization of vitreous approximately 900 µm above the retina in the foveal centre, and reported a sensitivity of 71% and a specificity of 88% [11]. This is better than our study, with a presented sensitivity of 54.8% and a specificity of 85.9%. Poor sensitivity in this study is consequent to a relatively high number of false positive cases, which may be the result of the shallower section of vitreous included in our SD-OCT images. However, our results are significantly better than those reported by Stavrakas et al. with a sensitivity of 37.5% and specificity of 31.3%, which also included only macular cube images [14]. Interestingly, three dimensional OCT images of the optic disc in a relatively small sample of patients provided the highest sensitivity of 97.4% with a reported specificity of 100% [10].

On the other hand, vitreoretinal adhesion or separation in the optic nerve can be inferred on a 6-mm macular OCT scan, although the optic nerve is not captured in the scan area [34]. Like previous research, we assumed that if the posterior vitreous cortex was visible on the top part of the scan, it was still attached to the optic nerve. On the contrary, we assumed that if the posterior vitreous cortex could not be visualized on the 6-mm OCT scan, it was separated from the optic nerve, that is, that a complete PVD was present. Therefore, 6-mm scans may be sufficient to assess PVD status, and visualization of the vitreoretinal interface at the optic nerve may be less important than previously claimed as long as the images are not shifted too far superiorly [34]. In this research, we insisted on high-quality images of the retina with the lowest possible position of the retinal image.

Optical coherence tomography imaging is an emerging standard of care in the setting of patients presenting with new flashes and floaters [42]. Enhanced depth imaging (EDI) for SD-OCT has further improved depth sensitivity, which enables detailed monitoring of choroidal structure and measurement of choroidal thickness [43]. Recent development of swept-source OCT (SS-OCT) has enabled imaging of larger areas with reduced motion artifact, and a better visualization of the choroidal vasculature [44]. The use of wide-field SS-OCT technology, which allows simultaneous observation of the optic nerve and macula, along with a greater depth of the vitreous included in the OCT image, will certainly enable a more reliable assessment of PVD state in the future [46]. More recent advances include a wide variety of new technologies, including wide-field colour photographs, quantitative fundus autofluorescence (qAF), adaptive optics (AO), and fluorescence lifetime imaging ophthalmoscopy (FLIO) [45]. However, it should be kept in mind that each technology has different capabilities and, to date there is no single modality capable of providing all the necessary information for a certain disease.

In this study, we have also analysed the combined value of all three diagnostic methods. The combined use of SD-OCT and US provided the best results, while the addition of BM did not provide a significant improvement. The rationale is simple, while SD-OCT has the ability to capture shallow PVD, US, on the other side can detect PVD that is highly detached over the retina and also visualize the optic nerve head. Furthermore, our findings indicate that the age and sex of patients, which are highly predictive of PVD status, when considered jointly in the models, significantly improve the diagnostic accuracy of the three diagnostic methods alone.

The strengths of this study include the prospective design and the availability of a wide range of preoperative and intraoperative details. We have insistently used only high-quality images and used the same preoperative and intraoperative protocol, including ILM peeling, in all our patients. The limitations of this study include the examiner related bias, with BM performed and interpreted by one examiner and US by another examiner, and SD-OCT and intraoperative findings analysed by both examiners. A possible source of error is a change in PVD status from examination to surgery [11, 16]. To minimize this possible bias, all patients included in this study underwent a complete standardized examination one day before the surgery [9, 14].

In conclusion, we present the comparative findings of three diagnostic procedures with the intraoperative findings of PPV. Preoperative BM, US, and SD-OCT are associated with relatively poor sensitivity and relatively good specificity for detecting complete PVD. A combination of all three diagnostic methods can provide a good insight into the condition of the vitreoretinal interface. A detailed preoperative examination and planning are necessary to achieve the best intraoperative and therefore postoperative results.