Prediction of long-term prognosis of age-related macular degeneration treated by hemorheologic therapy using baseline laboratory indicators - Experimental-clinical model

1.1 Characteristics of AMD and therapy possibilities

AMD is a degenerative retinal disease that causes progressive changes of visual acuity (visus) in patients and is the most common cause of practical blindness in people over 60 years of age in industrialised countries [1]. The incidence of this disease increases with increasing average age of the population. Some new information regarding diagnostics and therapy is stimulating, AMD, however, continues to be an incurable disease. Serious loss of visual acuity occurs in approximately 10–15% of patients [2–5].

Despite all efforts, there is currently no satisfactory therapy for AMD. Treatment modalities include changes of lifestyle (no smoking, physical activity, omega-3 fatty acids), and increased use of antioxidants (vitamin C, vitamin E, β-carotene, zinc and copper). Based on the AREDS2 study, the aforementioned combination of antioxidants had a partial effect on reducing the risk of AMD progression –after 5 years, progression occurred in 20% and 28% of patients in the treated and placebo groups respectively [6]. Other treatment options are subject to ongoing research –complement inhibitors, prostaglandins, anti-amyloid antibodies, neuroprotective medication, and others [1, 7].

Another treatment modality in the case of the dry form of AMD is rheohaemapheresis, which we have been developing in our Department over a long period [8–10]. Rheohaemapheresis leads to elimination of a precisely defined spectrum of high-molecular weight proteins, including fibrinogen, 2-macroglobulin, LDL-cholesterol, IgM, fibronectin, and von-Willebrand factor. These changes improve erythrocyte flexibility and result in the reduction of whole blood and plasma viscosity, erythrocyte and thrombocyte aggregation. Repeated pulses of lowered whole blood and plasma viscosity lead to rapid changes of blood flow, subsequently inducing sustained improvement of the microcirculation [11, 12]. In this context, it represents the complete interactive network between epithelium, fibroblasts, and endothelial cells, extracellular matrix, and blood components [13]. Repeated results show that rheohaemapheresis may lead to the reduction of AMD progression in the long-term perspective [14, 15]. Rheohaemapheresis in the dry form of AMD is also recommended by the Guidelines of the American Society for Hemapheresis [16].

3.1 Rheohaemapheresis method used as treatment

We used our own modification of Borberg’s “rheohaemapheresis” [17–19]. To obtain plasma, we used continuous separators (Cobe Spectra or Spectra Optia, Terumo BCT, Lakewood, Co, USA) and Evaflux 4A filters (Kawasumi, Tokyo, Japan) to wash the obtained plasma. The flow through the filter was controlled, using the CF100 automatic machine (Infomed, Geneva, Switzerland). See Fig. 1. Anticoagulation was performed using a combination of heparin and ACD-A (Baxter, Munich, Germany). Between 100% and 150% of body blood volume was washed. The procedures were performed from the peripheral vein in the elbow pit or in the forearm. We presented some more detailed data separately [20, 21].

OPEN IN VIEWER

Fig. 1

Principle of Rheohaemapheresis. Legend: Blood is collected from the peripheral vein, plasma is obtained in the separator and flows through the Evaflux filter. Washed plasma with erythroctes is then returned to another peripheral vein of the patient.

The treatment scheme was performed for patients with AMD according to the MIRA-1 study [22]; 8 procedures in 10 weeks, i.e. 2 aphereses weekly at intervals of 2–4 days, followed by a break of 14 days, followed by another series of aphereses. The patients were then followed up every 6 months. The effects of a successful treatment last from 2 to 2.5 years, according to our experience. It is recommended (based on the literature data and our experience [13, 23]) that 1–2 additional procedures are performed to boost the effect (“booster therapy”) after this 2-year period. For our group of patients, we decided on 2 procedures within 1 week.

3.3 Patients and methods

We treated 81 patients with AMD, using rheohaemapheresis between January 2006 and August 2019. The patients were followed up for a long period of time. As of 31 August 2019, the median of follow-up was 8.95 years. In this study, 66 patients who completed rheohaemapheresis treatment using the predetermined scheme, with at least a 5-year follow-up, were included for evaluation. The group consisted of 28 males and 38 females. Average age was 75.2±6.6 years.

The patients were divided into 2 groups. The first group consisted of patients who were clinically successfully treated, and the second group included patients with therapeutic failure. Out of 66 patients, 52 were evaluated as successfully treated, 14 patients as treatment failure. Patients whose disease became stable or improved (visual acuity, morphological ocular findings) were considered as being successfully treated. (Note: the group of successfully treated patients is numerically larger, as it includes patients who were found to have improved, as well as those with a stable disease. Generally, long-term stabilisation is considered as successful therapy in the case of AMD). The most important criterion, especially from the patient’s point of view, is visual acuity: this was evaluated as stabilisation if the visual acuity was within±10 letters of the original value on a specified date (i.e. change of maximum of 2 lines of the ETDRS optotypes). Improvement of visual acuity by 11 or more letters was evaluated as an improvement; deterioration by 11 or more letters was evaluated as a worsening. Stabilisation of the morphological finding was evaluated as a change in the original area of pathological changes (drusenoid pigment epithelium detachment, area of soft drusen, area of RPE-atrophy) by±25%. A decrease of more than 25% in the pathological changes was considered as an improvement; analogical increase by more than 25% from the baseline was considered as a worsening. The effect of the therapy was evaluated by an experienced ophthalmologist.

All patients signed informed consents and the study followed the guidelines of the Helsinki Declaration. Study was also reviewed and approved by University Hospital Hradec Kralove Ethics Committee (approval number 201607 S03P).

4.1 Prognosis prediction using rheologically significant biomarkers

The results of the following examinations were used for comparison:

1) Rheologically effective parameters

Spectrum of rheologically effective high-molecular proteins according to Klingel et al. [13]:

Viscosity of whole blood and plasma.

2) Special clinical examinations used for evaluation of the therapy

Anatomical changes were evaluated every 6 months: the area of soft drusen, the area of retinal pigment epithelium detachment, the area of RPE atrophy in a colour photo using the VISUPAC program (fundus camera FF450 + IR, Zeiss, Germany); the presence of ellipsoid layer defects in the transfoveolar scan by optical coherence tomography (Cirrus, Zeiss, Germany). We also reviewed changes in the retinal functional status of patients after the rheohaemapheresis treatment: the best corrected visual acuity on ETDRS boards and electrophysiological examinations of retinal neuron function: electroretinography (ERG), ERG and multifocal ERG (Electrophysiological System+mf ERG; Roland Consult, Germany).

Evaluation of therapy success must be based on a combination of all the special examinations stated above. However, the most important criteria (from the patient’s point of view) is visual acuity. Specific details were not part of this study.

We compared the levels of rheologically active factors in the group of clinically successfully treated patients (52 patients) versus the unsuccessful patients (14 patients) (see Table 1). There was no statistically significant difference in age (p = 0.349) or gender (p = 0.698) between the two groups. We performed this evaluation, using paired samples taken from the patients at the beginning of a series of rheohaemapheresis. One sample was taken before and one after the rheohaemapheresis. Lipoprotein(a) was evaluated only during examinations performed before the laboratory limits in our laboratory changed (September 2014); a smaller number of samples was also used for the examination of immunoglobulin M.

The results show that the levels of rheologically significant factors evaluated before and after the treatment were not significantly different between the groups of clinically successfully treated patients and those with treatment failure. The prognosis therefore cannot be determined by using the baseline levels of individual rheologically important factors. See Table 1.

We also took and compared samples at the end of the whole treatment cycle with equivalent results (data not presented).

Other individual laboratory indicators were analysed, using classic statistical methods (t-tests, ANOVA, Pearson correlation analysis). We carried out the following: parameters of blood count, basic coagulation examinations (aPTT, INR), basic biochemical examinations (blood minerals, blood proteins, creatinine, glycaemia, liver enzymes, uric acid). There were no significant differences between the baseline and outcome results for the two groups of patients (successfully/unsuccessfully treated).

4.3 Evaluation using the method of discriminant analysis

We chose multivariate statistical analysis, using discriminant analysis for further data analysis. We used Systat 13 software (Systat, Chicago, IL, USA) for analysis of the linear and quadratic models (using the automated and interactive step-wise approach).

Only those patients for whom the values of all the factors in the calculation were available were included in the calculation of the classification function. Selection of the spectrum of anticipated effective factors: we selected a spectrum of rheologically active high-molecular proteins according to Klingel et al. [13], i.e. total cholesterol, LDL-cholesterol, apolipoprotein B, IgM, 2-macroglobulin, fibrinogen, whole blood and plasma viscosity. Other factors, called “secondary factors”, were also selected –indicators which are readily available and commonly used: albumin, total protein, haemoglobin, leukocytes, thrombocytes, creatinine, glycaemia (other factors could be included in the evaluation in future research).

The system determined the coefficients of the two classification functions for our model (one for successful and the other for unsuccessful treatment), gradually eliminating the insignificant indicators. The following indicators remained for further evaluation: glycaemia, creatinine, total cholesterol, LDL-cholesterol, lipoprotein(a), and whole blood and plasma viscosity. All patients were classified using this method. The program then performed a final evaluation and determined the percentage correctness of the prediction of successful treatment. In this case, the prediction was correct in 79% of patients with unsuccessful treatment and in 60% of patients with successful treatment. The overall prediction correctness was 64%. See Table 2 - upper part. We also used the “Jackknife” method. However, this provided less optimistic results (57% and 54% of predictions were correct). See Table 2 - lower part.

The results were more accurate when the quadratic model was used. In this case, the prediction was correct in 100% of unsuccessfully treated patients and in 75% of successfully treated patients, 80% for all patients overall (66 patients were evaluated in this way). For the results of the final evaluation of the program, see Table 3.

The program for the prediction of prognosis, using the model described in Table 2 as well as the quadratic model –see the classification matrix in Table 3 –is currently being run in a trial version at https://www.vejrazka.name/apps/med/prognoza/, but there is no guarantee that it will be available at any particular time in the future.

The results of the prediction of prognosis using the stated experimental model are encouraging. They show some optimistic tendencies in the prediction of prognoses. However, the number of patients in our study is still insufficient to draw convincing conclusions. The compilation of this model, however, allows the possibility to continue relatively easily during further testing –by gradually collecting further results obtained during ongoing research activities and recruitment of further patients. After inclusion of the results in the created matrix (see above - in a trial version on https://www.vejrazka.name/apps/med/prognoza/), the computer automatically calculates the probable prognosis of the patient –the probability of successful or unsuccessful therapy is predicted. It would then be possible to choose a more intensive procedure at the beginning of treatment, based on the result. The laboratory values that are used in our model are relatively easily available. Once the patient is recruited to the research protocol and undergoes the first rheohaemapheresis, the laboratory results in question are available. It will be possible to re-evaluate the model more accurately and to specify its practical value when data from the necessary number of patients are available.

5.2 Further research directions

To broaden the focus of our research, we have also investigated the dynamics of plasma circulating biomarkers (microRNAs) during the follow-up of AMD patients. MicroRNAs are small RNAs (18–22nt) that play an important role in the regulation of gene expression. We found significant changes in miRNAs levels in response to rheohaemapheresis treatment in our research cohort of 9 patients (Dlouhá et al, unpublished data). It is possible that the differences in occurrences of certain types of miRNAs are another suitable marker for assessing the long-term prognosis, as is the case of some other diseases. This is the subject of our ongoing research.