Autologous Serum Eye Drops for Dry Eye: Systematic Review : Optometry and Vision Science (2024)

Dry eye is “a multifactorial disease of the ocular surface characterized by a loss in homeostasis of the tear film and accompanied by ocular surface symptoms, in which tear film instability and hyper-osmolarity, ocular surface inflammation and damage, and neuro-sensory abnormalities play etiological roles.”¹ Dry eye is common, with an estimated prevalence of 5 to 50%, depending on population characteristics including age and sex.² Based on a 2022 systematic review, the estimated prevalence in the United States is 8.1% (95% confidence interval [CI], 4.9 to 13.1%).³ No criterion standard diagnostic test for dry eye exists. The main methods of assessment were using patient-reported symptoms from questionnaires and objective clinical measurements of the eye and tear film, which have inconsistent overlap.^4–8

Many systemic diseases are associated with dry eye. A common condition known to cause dry eye is Sjögren syndrome.^9,10 It is estimated that 1 in 10 patients with clinically significant dry eye has underlying Sjögren syndrome.^11,12 Other potential causes of dry eye include drug adverse effects (e.g., retinoic acid therapy),¹³ laser in situ keratomileusis surgery,¹⁴ cataract surgery, and contact lens use.¹⁵

Treatments for dry eye include artificial tears, topical cyclosporin A, topical steroids, punctal plugs, nutritional supplements, intense pulsed light, and others. The effectiveness and safety of individual treatments have been summarized in several Cochrane systematic reviews.^16–20 A commonality with all of the aforementioned treatments is that they do not reflect the composition of natural tears, which contain various lipids, carbohydrates, proteins, vitamins, and growth factors.²¹ To better approximate natural tears, eye drops produced from blood serum have been investigated and are currently used in some cases of dry eye. Both autologous and allogeneic serum eye drops have been studied, with allogeneic serum posing safety concerns regarding hypersensitivity reactions and transmission of blood-borne diseases.²² For these reasons, autologous serum has been preferred over allogeneic serum when possible.²² The objective of this systematic review is to assess the effectiveness and safety of autologous serum eye drops for dry eye.^23,24

METHODS

Eligibility Criteria

We included randomized controlled trials (RCTs) of dry eye patients 18 years or older. We included trials that had compared autologous serum with or without artificial tears versus artificial tears, saline, or placebo alone or with no treatment.

Outcomes of Interest

The critical outcomes of this review were study-defined subjective symptoms, severity, and frequency reported on validated measurement scales (e.g., Ocular Surface Disease Index [OSDI], visual analog pain scale). Other important outcomes included tear osmolarity, corneal stain, conjunctival stain, tear film breakup time (TBUT), Schirmer test, corneal topography, impression cytology, conjunctival biopsy, quality-of-life measures, and economic outcomes. All these outcomes were analyzed as mean change from baseline to up to 4 weeks' follow-up. We also assessed the risk of adverse events during the same period.

Search Methods

To identify trials of potential relevance for inclusion in this review, we searched CENTRAL, Ovid MEDLINE, EMBASE, Latin American and Caribbean Literature on Health Sciences, ISRCTN registry, ClinicalTrials.gov, and the World Health Organization International Clinical Trials Registry Platform from their inception to September 30, 2022 (full search criteria in Appendix Tables A1 and A2, available at https://links.lww.com/OPX/A612). We also searched citations of included trials for potential additions.

Study Selection

Two authors independently screened the titles and abstracts yielded by the searches and then judged study eligibility based on relevant full-text reports. A third review author resolved discrepancies. We contacted study authors via e-mail whenever necessary to judge eligibility. If we received no response after 2 weeks, we labeled the study as “awaiting classification” to indicate that we were unable to determine eligibility with information available. We recorded the reasons for exclusion of full-text reports (Appendix Table A3, available at https://links.lww.com/OPX/A612).

Data Extraction and Risk-of-bias Assessment

We merged multiple reports (e.g., trial registration, journal publication) of the same study before data extraction. Two authors independently extracted relevant study characteristics, methods, and results from each included trial using a previously piloted data extraction form in Covidence.²⁵ Two authors followed the same steps to assess potential sources of bias using the Cochrane Risk of Bias tool, with each domain judged to confer “high,” “low,” or “unclear” risk of bias.^26,27 Disagreements between review authors during data extraction and risk-of-bias assessment were resolved by discussion. We requested information from trial investigators when study design or outcome data were unclearly reported. If we received no response after 2 weeks, we used the available data.

Data Synthesis

We calculated the mean difference (MD) and 95% CI for continuous outcomes. All analyses were performed in RevMan Web.²⁸ We compared clinical heterogeneity (participants, interventions, outcomes) and methodological heterogeneity (study design, analysis, and risk of bias) to determine whether studies were similar enough to produce an informative meta-analysis. We reported the data in a narrative format when this was not possible. We assessed the certainty of evidence as “very low,” “low,” “moderate,” or “high” based on the Grading of Recommendations Assessment, Development and Evaluation approach.²⁹ We downgraded from high-certainty evidence, because of the randomized design of included studies, based on four main criteria: risk of bias, inconsistency, indirectness, and imprecision.²⁹

RESULTS

The electronic searches yielded 929 titles and abstracts. After screening the full text of 37 studies (40 reports), we included 6 studies (9 reports; Fig. 1), listed 1 study report as “awaiting classification,” and provided reasons for the 30 excluded studies (Appendix Table A3, available at https://links.lww.com/OPX/A612).

Description of Included Studies

The six included RCTs are summarized in Table 1. These trials were conducted in Japan (two studies),^14,30 Turkey (two studies),^13,31 Australia (one study),³³ and Chile (one study)³² and had enrolled a total of 115 participants. Participant numbers per trial ranged from 12 to 27. The average age of participants ranged from 25 to 64 years. A total of 72 women and 43 men were enrolled.

Vials of autologous serum were stored in the freezer and then moved to the refrigerator during use by participants for all the studies except for Yilmaz et al.,¹³ where participants were instructed to refrigerate all eye drop containers for the 4-week treatment phase. Five trials used 20% autologous serum as the intervention, but Yilmaz et al.¹³ used 40% autologous serum.

Of the two trials that had compared autologous serum with saline, there was one trial designed as parallel group,¹⁴ and one had a paired-eye design.³³ Of the four trials that had compared autologous serum with artificial tears, one used a parallel-group design,³⁰ and three used a crossover design.^13,31,32 All three crossover trials included a washout period to minimize carryover effects, and one also had a 2-week washout period before beginning the trial.³¹

No trial was judged to have low risk of bias across all domains (Fig. 2). We judged four of the six trials to have had unclear or high risk of bias in five or more of the nine domains assessed. All six of the trials were judged to have had high or unclear risk of bias for the selective reporting domain.

We did not conduct any meta-analysis because the outcomes reported were diverse and variable across included studies. Furthermore, the analyses reported from crossover trials seemed to be inappropriate. In the following sections, we describe available results by comparator.

Autologous Serum Compared with Artificial Tears

Of the three crossover trials that compared autologous serum with preservative-free artificial tears, two tested 20%^13,31,32 autologous serum and one tested a 40% concentration.¹³ In addition, one parallel-group trial also compared 20% autologous serum with artificial tears.³⁰ The outcomes and certainty of evidence are shown in Table 2.

In terms of patient-reported symptoms, one trial reported the mean 2-week change and standard deviation from baseline as measured on a visual analog pain scale (0 to 100, with 100 indicating “unbearable pain”). The change values were −19.2 ± 8.8 for the 20% autologous serum group and −7.2 ± 9.8 for the artificial tears group, with an MD of −12.00 in favor of autologous serum (95% CI, −20.16 to −3.84; 20 participants [37 eyes]; Fig. 3; low-certainty evidence).³⁰ Two crossover trials also assessed participant-reported symptoms after 4 weeks of treatment with the OSDI questionnaire.^13,31 Both studies reported decreased OSDI scores, indicating improvement, in the autologous serum groups versus the artificial tear groups. However, no information was provided about sample size per sequence initially or at follow-up time points, making it impossible for us to estimate any effect. Furthermore, the reported analyses did not mention having accounted for the correlations because of pairing and repeated measures inherent in trials with crossover designs.

Three trials reported data on fluorescein staining. One trial reported an MD from baseline to 2 weeks' follow-up of −0.90 in favor of autologous serum (95% CI, −1.47 to −0.33; 20 participants [37 eyes]; Appendix Fig. A1, available at https://links.lww.com/OPX/A613; low-certainty of evidence).³⁰ This difference is small and may not be clinically important. The other two trials reported small differences between groups in mean Oxford Scale scores for fluorescein staining^31,32 that were neither statistically nor clinically significant. One trial reported a possible clinically significant difference between 20% autologous serum and artificial tears on Rose Bengal staining, with an MD from baseline to 2 weeks' follow-up of −2.20 in favor of autologous serum (95% CI, −2.73 to −1.67; 20 participants [37 eyes]; Appendix Fig. A2, available at https://links.lww.com/OPX/A614; low-certainty evidence).³⁰ However, this difference may not indicate a clinically meaningful improvement because both groups had scores considered to be abnormal (score >1) at both baseline and follow-up.³⁰ Finally, the registration record of one trial mentioned corneal and conjunctival staining but did not specify the staining technique and did not report results.¹³

Four trials reported data on TBUT. One trial reported that the difference between 20% autologous serum and artificial tears was not clinically significant, with an MD from baseline to 2 weeks' follow-up of 2.00 seconds in favor of autologous serum (95% CI, 0.99 to 3.01 seconds; 1 RCT, 20 participants [37 eyes]; Appendix Fig. A3, available at https://links.lww.com/OPX/A615; low-certainty of evidence).³⁰ Two trials reported differences in TBUT between groups as 1 and 2 seconds.^31,32 One trial reported median and standard deviation (or standard error) TBUT values for each period by intervention sequence before and at the end of 1 month.¹³ The investigators of these three crossover trials seem not to have accounted for the crossover design and did not provide sufficient information for comparison of scores between treatment groups.

Two trials reported data on the Schirmer test. One trial assessed the Schirmer test without anesthesia, finding that both groups had severe dry eye (<4 mm) at 2 weeks after the start of treatment, with a between-group MD of −0.40 mm (95% CI, −2.91 to 2.11 mm; 20 participants; Appendix Fig. A4, available at https://links.lww.com/OPX/A616; low-certainty evidence).³⁰ Two trials reported no evidence of difference in Schirmer test scores between autologous serum and artificial tears.^13,31

No trial reported data on tear hyperosmolarity, corneal topography, impression cytology, conjunctival biopsy, quality-of-life measures, economic outcomes, or adverse events.

Autologous Serum Compared with Saline

Two trials compared 20% autologous serum with saline eye drops.^14,33 The outcomes and certainty of evidence are presented in Table 3. As shown, neither trial found any statistically significant difference for participant-reported symptoms, fluorescein staining scores, Rose Bengal staining, TBUT, or Schirmer test. One trial reported that the difference in Rose Bengal staining scores was not clinically significant, with an MD from baseline to 4 weeks' follow-up of −0.60 on a 0- to 9-point scale in favor of autologous serum (95% CI, −1.11 to −0.09; 35 eyes; Fig. 4; very low-certainty evidence). Neither trial reported data on tear hyperosmolarity, corneal topography, impression cytology, conjunctival biopsy, quality-of-life measures, economic outcomes, or adverse events.

DISCUSSION

The current systematic review revealed low-certainty evidence that autologous serum, as compared with artificial tears, may improve patient-reported symptoms after 2 weeks of treatment. In addition, very low-certainty evidence suggests that autologous serum, as compared with saline, may improve Rose Bengal staining scores after 4 weeks of treatment; however, no statistically significant difference was found for other objective clinical measures of the ocular surface status or patient-reported symptoms. Thus, there is insufficient evidence available to support the use of autologous serum for treating dry eye.

In 2020, the American Academy of Ophthalmology published a narrative review that suggested that autologous serum eye drops may be effective in treating severe dry eye, but the authors noted that there were not yet enough controlled trials with similar study protocols to reach more definite conclusions.³⁴ The authors of the American Academy of Ophthalmology report had classified evidence levels of included trials as levels “II” and “III”; they incorporated findings and conclusions from two of the six RCTs in our review. They examined persistent corneal epithelial defects in addition to severe dry eye; in contrast, in the current review, we focused only on dry eye.

Our findings are largely congruent with two recently published systematic reviews, Franchini et al.³⁵ and Wang et al.³⁶ The scope of Franchini et al. was ocular surface diseases, and only the meta-analysis portion specifically focused on dry eye. Franchini et al. included all six trials included in our review and concluded that the efficacy of autologous serum eye drops for dry eye could not be evaluated because of heterogeneity in study designs and because several trials had applied additional local therapies alongside autologous serum. Wang et al. included five of the six trials in our review and concluded that autologous serum eye drops may provide some benefits regarding the OSDI, TBUT, and Rose Bengal staining, although differing classifications of dry eye disease and concentrations of autologous serum were main factors that limited stronger conclusions. Although both Franchini et al. and Wang et al. included Mukhopadhyay et al.,³⁷ who studied patients with Hansen disease, we did not because we considered these patients' symptoms to be due to neurotrophic keratitis rather than dry eye disease. In addition, we chose not to conduct meta-analyses for the current review because of a high degree of unreliability in reported trial data, especially from crossover trials. Several assumptions would have been required to perform the transformations and imputations necessary for a formal meta-analysis. Ultimately, we concluded that these assumptions were too substantial to carry out a meaningful quantitative synthesis.

The strength of our review is in the steps taken to minimize bias and our up-to-date literature search. We have performed a comprehensive search, followed by screening, data extraction, risk of bias, and Grading of Recommendations Assessment, Development and Evaluation assessment in duplicate. We sought missing information from investigators and did not extrapolate findings. Despite these efforts, the limitations of the included studies prevented us from drawing firm conclusions. For example, the included trials enrolled only participants with severe dry eye disease. Many of the outcomes we wished to evaluate were not fully defined or were not reported at all.

Two concentrations of autologous serum (20 and 40%) have been studied in the trials included in our review. More research is needed to determine optimal and standardized methods for production and storage of autologous serum eye drops as well as the effects of varying autologous serum eye drop concentrations. The preparation of blood-based products requires specialized pharmacies that are not widely accessible. There are barriers associated with the drawing blood and with preparation and administration of blood-based products at specialized pharmacies. In the United States, insurance companies often do not cover blood-based products, and the out-of-pocket cost can range from $175 to $250 for a 2-month supply.³⁸ Even established efficacious treatments can lead to inequities in health care when such barriers are not considered when reporting findings from studies and reviews.

Development and utilization of a core outcome set for dry eye disease are essential to the future of evidence-based ophthalmology practice.^39–41 Future studies should use standardized and validated scoring systems of dry eye clinical severity. Objective biomarkers, such as tear osmolarity and tear cytokines, which are used as a parallel index to dry eye severity scales, should be applied as outcomes in conjunction with clinical assessments and participant-reported symptoms. Data on outcomes, especially adverse outcomes, should be collected over longer periods. A survey of dry eye participants indicated a preference for outcomes measured within 6 months.⁴¹ We found registration records for only two of the six included studies; neither of the two trials had been registered before enrolling participants, and neither had reported results on the registry record. Finally, for trials with crossover and within-person designs, registration, analysis of outcomes, and reporting should account for the paired design.^42–44

This review provides an update to a 2017 Cochrane systematic review on autologous serum eye drops for dry eye by adding one new trial for inclusion.²⁴ Despite the interest in autologous serum eye drops, there has not been overwhelming progress to further understand this treatment.

In conclusion, the effectiveness of autologous serum eye drops for the treatment of dry eye is unknown based on current data, and there is insufficient evidence to supporting their use.

REFERENCES

Supplemental Digital Content

• OPX_100_8_2023_08_16_QUAN_ovs23048_SDC1.pdf; [PDF] (212 KB)
• OPX_100_7_2023_06_30_QUAN_OVS23048_SDC2.tif; [Other] (210 KB)
• OPX_100_7_2023_06_30_QUAN_OVS23048_SDC3.tif; [Other] (210 KB)
• OPX_100_7_2023_06_30_QUAN_OVS23048_SDC4.tif; [Other] (179 KB)
• OPX_100_7_2023_06_30_QUAN_OVS23048_SDC5.tif; [Other] (179 KB)