Description of the condition
Dry eye is a common disorder, with an estimated 25% of patients in general ophthalmology or optometry clinics reporting dry eye symptoms (Doughty 1997). It is known that the incidence of dry eye increases with age and has a higher prevalence in women compared to men (McCarty 1998; Schaumberg 2003; Stern 2004). Recently, the Definition and Classification Subcommittee of the International Dry Eye Work Shop (DEWS), redefined dry eye as “a multifactorial disease of the tears and ocular surface that results in symptoms of discomfort (including foreign body sensation, dryness or irritation, burning, light sensitivity, redness), visual disturbance, secretion with crusting on the eyelashes, and tear film instability with potential damage to the ocular surface. It is accompanied by increased osmolarity of the tear film and inflammation of the ocular surface" (DEWS 2007). An increased tear osmolarity, which causes ocular surface inflammation, is thought to be the central pathogenic mechanism of dry eye (DEWS 2007).
The mechanistic classification of dry eye suggested by the DEWS defines two main subtypes: aqueous deficiency and evaporative dry eye, respectively corresponding to disorders of the lacrimal and meibomian glands (DEWS 2007). Disorders of the lacrimal and meibomian glands are usually secondary to either systemic diseases or local causes. One of the most common systemic diseases causing dry eye is Sjögren’s syndrome. It presents as “sicca complex,” a combination of dry eye and dry mouth (xerostomia) due to a T-lymphocyte mediated destruction of the exocrine glands (Fox 2006; Kumar 2005; Yamada 1990). Other systemic diseases, such as rheumatoid arthritis, diabetes, systemic lupus erythematosus, dermatological conditions such as acne rosacea, and Graves' disease have also been reported as causing clinically significant dry eye (Patel 2002). On the other hand, the leading causes for non–systemic disease-related dry eye include age-related lacrimal dysfunction (Demato 1984), hormonal changes, drug side-effects (e.g. systemic antihistamines, diuretics, topical beta blockers for glaucoma therapies) (Blomquist 2010; Baudouin 2001), surgical intervention (e.g. photorefractive keratectomy [PRK] and laser in situ keratomileusis [LASIK] (Campos 1992; Noda-Tsuruya 2006; Toda 2004), as well as long-term contact lens use (Lemp 1995).
The diagnosis of dry eye is made by validated patient symptom questionnaires and with a wide array of clinical assessments of the tears and ocular surface. Symptoms of dry eye have been standardized by use of questionnaires. The most common complaints described by patients include dryness or irritation, light sensitivity, foreign body sensation, red eyes, poor vision, daily life limitations, and symptom fluctuation in different environmental conditions. However, it has also been noted that there is no strong correlation between signs and symptoms, particularly in mild dry eye (Begley 2003; Viso 2012). Therefore, the clinical diagnosis of dry eye needs to incorporate objective tests such as tear osmolarity, tear production by Schirmer's testing, fluorescein clearance, fluorescein break-up time (BUT), and demonstration of ocular surface damage through dye staining (fluorescein and lissamine green) (Lemp 2011; Lemp 1995; Perry 2004). Although there is presently no gold standard diagnostic test to identify dry eye, a growing number of studies have suggested that tear osmolarity might be the best single metric for diagnosis and severity assessment of dry eye (Lemp 2011; Tomlinson 2006). According to Perry 2004, other authors also suggest tear film stability by BUT and delayed tear fluorescein clearance (Chodosh 1994; Marci 2000) as reliable ways to assess dry eye.
Currently there is no cure for dry eye. Common treatments are targeted to manage the symptoms. The mainstay of conventional therapy is the application of artificial tears that increase moisture on the ocular surface, and provide additional lubrication. A variety of artificial tear formulations differ from each other in their electrolyte composition, osmolarity, viscosity, the presence of preservatives, and compatible solutes (Lemp 2008). An unpreserved artificial tear containing 0.1% sodium hyaluronate was found to be effective in improving dry eye symptoms with a significant improvement in the mean tear film osmolarity, break-up times, and conjunctival and corneal staining scores (Nelson 1988). However, the use of artificial tears has some limitations. Natural tears have a complex composition of water, salts, hydrocarbons, proteins, and lipids that artificial tears cannot exactly substitute (Dogru 2011; Quinto 2008). Additionally, frequent application of artificial tears solutions containing chemical preservatives to prevent contamination has been found to induce toxic and allergic reactions, especially among those with sensitive eyes (Baudouin 2010; Dogru 2011; Quinto 2008).
Topical corticosteroids that target the inflammatory pathways associated with ocular inflammation have been shown to improve symptoms in people with dry eye (de Paiva 2008; Pflugfelder 2004), but their use is limited due to long-term side effects including cataracts and increased intraocular pressure (Blomquist 2010). In December 2002, the U.S. Food and Drug Administration (FDA) approved 0.05% solution of cyclosporine A (CsA) as an ocular therapeutic for people with dry eye (Meadows 2005). Several studies have shown an increase in tear production and conjunctival goblet cell density with few reported adverse effects following the topical application of CsA (Sall 2000; Stevenson 2000; Toker 2010; Wilson 2007).
Additional nutritional supplements such as essential fatty acids, including omega-3, linoleic acid, and gamma-linoleic acid, have been proposed as adjuvants in the treatment of dry eye due to their anti-inflammatory properties (de Paiva 2008). Increased water intake and reduced alcohol consumption are also recommended to improve dry eye symptoms (Dogru 2011). Environmental interventions designed to increase air moisture and reduce particles in the air, including indoor humidifiers and air filters or cleaners have been shown to reduce dry eye symptoms as well (Dogru 2011). For people in whom artificial tears are not sufficient, preservation of the tear film can be achieved by inserting punctal plugs in the lacrimal ducts, designed to reduce the drainage of tears through the lacrimal ducts and increase lubrication on the ocular surface (Ervin 2010; Foulks 2003).
Description of the intervention
The composition of serum resembles that of tears; most concentrations are equivalent, with the exception of more vitamin A, lysozyme, transforming growth factor-β (TGF-β) and fibronectin, and less immunoglobulin A (IgA), epithelial growth factor (EGF) and vitamin C in serum than in tears (Bradley 2008; Joh 1986; Matsumoto 2004; Nelson 1992; Tsubota 1999). Since many of the essential components in tears are present in serum, the use of serum as a tear substitute for the maintenance of the ocular surface seems feasible (Imanishi 2000; Kojima 2005b). In 1975, autologous serum eye drops (AS) were initially applied for dry eye and reported by Ralph 1975. Since then, AS have become increasingly popular for treating ocular surface diseases, mainly dry eye.
Production of autologous serum eye drops
Currently, there are no commercially available forms of AS; AS must be compounded using autologous serum. Technological factors affect the product quality and properties of AS (Geerling 2004; Liu 2005). Even though there is large variability in the methodology for AS preparation, storage and administration, standards have been established to optimize therapeutic effectiveness and product safety (Geerling 2004; Liu 2005). In brief, blood is first drawn from the recipient and allowed to clot in the absence of an anticoagulant. Once a clot has formed, the supernatant is centrifuged to separate the serum from the solid components without inducing hemolysis. After centrifugation the serum is decanted into a sterile container and then may be diluted to the desired concentration. Autologous serum typically is administered in 20% concentration which is based on the concentration of the biological factors in actual tears, although higher concentrations (between 50% and 100%) have been used (Dogru 2011; Geerling 2004; Kojima 2008; Quinto 2008). There is always the possibility that serum may contain components that are detrimental to the ocular surface. TGF-β, for example, is known to have antiproliferative effects, and high concentrations of TGF-β may suppress wound healing of the ocular surface epithelium (Tsubota 2000). This was one of the reasons for using a diluted solution of serum in order to maintain TGF-β levels that are comparable with tears. Preservatives are usually not added to AS, thus reducing the risk of preservative-induced toxicity associated with other dry eye treatments. However, the lack of preservatives theoretically increases the risk of ocular infections. Autolgous serum can be stored for less than one month at 4°C while in use, and for up to three months at -20°C (Tsubota 1999). It is important that vials containing autologous serum be kept away from light to avoid degradation of vitamin A.
Autologous serum eye drops have been recommended for the treatment of several ocular surface disturbances, such as Sjögren’s syndrome-related tear deficiency, non-Sjögren’s tear deficiency associated with graft-versus-host disease, neurotrophic keratitis, persistent epithelial defects, superior limbic keratoconjunctivitis, and postoperative dry eye induced by LASIK. People who were treated with 20% to 50% AS four to eight times a day reported subjective improvement in dry eye symptoms; objective improvement based on fluorescein staining and break-up time tests also was observed (Chiang 2007; Hyon 2007; Kojima 2005b; Matsumoto 2004; Ogawa 2003; Poon 2001; Tananuvat 2001).
AS are usually well tolerated, and most recipients report improvement of discomfort. Occasionally, they may experience increased discomfort, slight epitheliopathy (drop-out of the corneal epithelial cells, akin to fluorescein staining of the surface of the eye), bacterial conjunctivitis, or eyelid eczema (Ogawa 2003; Rocha 2000; Tananuvat 2001). Fox 1984 reported no serious complications but mentioned that others had encountered scleral vasculitis and melting in people with rheumatoid arthritis. McDonnell 1988 described complications such as the deposit of immunoglobulins in the cornea and the presence of corneal peripheral infiltrates with 100% autologous serum treatment in one person.
Risk of infection
Because some of the serum’s components may have bacteriostatic effects, for example, lysozyme, complement, and immunoglobulin G (IgG), the addition of a further bacteriostatic agent may not be necessary. It is reported that AS can be used safely in an outpatient as well as inpatient setting, under a strict protocol of preparation and storage (Langnado 2004; Partal 2011). However, even though AS are prepared under sterile conditions on an individual patient basis, there are risks for contamination, and consequent infection, during the preparation, storage, and use of the drops (Geerling 2004; Lee 2008).
Selection of people suitable for autologous serum
In the United States, the FDA and the American Association of Blood Banks (AABB) have specified criteria for autologous blood donors, which include a minimum hemoglobin concentration of 11 g/dL (hematocrit of 33%) and deferral for conditions presenting risk of bacteremia. Additional criteria may be applied by the individual blood collection facilities and medical providers; these often specify that the patient must be well enough to undergo venipuncture several times a year and withstand loss of blood (Noble 2004; Roback 2008). Blood collection facilities sometimes specifically defer people considered to be at greatest risk from blood donation such as those with unstable angina, recent myocardial infarction or cerebrovascular accident, significant cardiac or pulmonary disease with chronic symptoms but who have not been evaluated by the treating physician, or untreated aortic stenosis. Children and pregnant women often are excluded (Roback 2008).
To prevent the risk of viral transmission to others (e.g. production or nursing staff and children at home who may unintentionally use serum eye drops), it is strongly recommended that the donor be tested for blood-transmitted diseases (e.g. human immunodeficiency virus (HIV), hepatitis B virus (HBV), hepatitis C virus (HCV), and syphilis), that hospital staff be cautious of serum production, and that the identity of the recipient be confirmed (Geerling 2004; Yoon 2007). Though there are significant legal ramifications due to the potential transmission of blood-based diseases to medical staff as well as serum recipients, there is no consensus as to whether people who have blood-transmissible diseases should be disqualified from donating serum for their personal use when medically indicated.
AS are unique among ophthalmic therapies in that they are manufactured specifically for each individual and are made from that person’s own blood. The regulations on autologous blood donation vary from country to country. In the United States the FDA’s Center for Biologics Evaluation and Research (CBER) is responsible for the regulation of blood intended for transfusion, blood components and derivatives. In the European Union (EU), several directives on AS have been issued (1965/65, 1975/139, 1975/318) by the European Parliament and Council. However, these directives had to be taken into account in the laws of each member state of the EU (Geerling 2004). For example, the National Blood Service in England and Wales has supplied AS under a drug exemption certificate for the purposes of a clinical trial from the regulatory body in the United Kingdom, the Medicines and Healthcare Regulatory Agency (Noble 2004). Special regulations by the FDA and other regulatory agencies for using blood products must be taken into account when considering the integration of AS therapy into treatment regimens for AS (Geerling 2008; Noble 2004; Roback 2008).
How the intervention might work
Studies have shown that AS contain biochemical factors such as EGF, vitamin A, TGF-β, fibronectin, substance P, insulin-like growth factor 1 (IGF-1), nerve growth factor (NGF), and other cytokines that are essential for the proliferation, differentiation, and maturation of the normal ocular surface epithelium (Gordon 1995; Matsumoto 2004; McCluskey 1987; Nishida 1983; Nishida 1987; Phan 1987; Poon 2001). Therefore, a potential advantage of AS over traditional therapies is that AS serves as a lacrimal substitute to provide lubrication and other biochemical components of tears to assist in corneal and conjunctival epithelium maintenance with limited toxicity (Dogru 2011; Geerling 2004; Liu 2005; Poon 2001; Quinto 2008).
Why it is important to do this review
The use of AS in severe dry eye treatment has gained widespread acceptance in the past decade. However, it continues to be a restricted area because the preparation of serum eye drops requires a well-equipped laboratory and trained personnel. Studies conducted recently are controversial with regard to the effectiveness of AS for dry eye symptoms (Noble 2004; Tananuvat 2001). Therefore, we undertook a systematic review to determine the efficacy and safety of AS for the treatment of dry eye.
The aim of this review was to evaluate the efficacy and safety of AS as compared to artificial tears in the treatment of dry eye in adults.
Criteria for considering studies for this review
Types of studies
We included only randomized controlled trials (RCTs) for the purpose of this review. Given the stability of the condition of interest, we also considered cross-over studies in which the sequence of treatments was determined to have been assigned randomly.
Types of participants
We included in the review studies conducted in adults (age over 18 years old), with dry eye defined by the study investigators with no restrictions based on race or sex.
Types of interventions
We included studies in which the application of AS alone or in combination with artificial tears was compared to artificial tears alone, saline, placebo, or no treatment.
Types of outcome measures
Dry eye clinical tests generally do not correlate with participant-reported symptoms. There are a wide variety of participant-reported outcome scales that actually lead to the discrepancies between subjective symptoms and objective clinical tests (Chambers 1999; Fuentes-Paez 2011; Nichols 2004; Patrick 2011). Therefore we took into consideration both subjective data from participant-reported symptoms regardless of measurement scale and objective data obtained from clinical diagnostic tests to analyze fully their effect on the condition.
We defined symptom improvement as the change from baseline in participant-reported severity and/or frequency of dry eye-related symptoms based on validated patient symptom questionnaires at four weeks after initiation of treatment. Since trial design, frequency of AS administration, and timing of outcome assessment may vary, we considered all variations in frequency of AS use and other time points as reported by included studies.
- Tear hyperosmolarity: mean change in tear osmolarity.
- Ocular staining with fluorescein: mean change in total score from baseline to follow-up.
- Tear film break-up time: mean change in tear film break-up time in seconds.
- Schirmer’s test: mean change in millimeters with or without anesthesia.
- Ocular staining with Rose Bengal: mean change in total score from baseline to follow-up.
- Corneal topography: mean change in tear film break-up time and the height of the tear meniscus by non-invasively assessing the tear film.
- Impression cytology: mean change in grades of epithelial metaplasia and goblet cell density.
- Tear fluorescein clearance: mean change in the speed of disappearance from the ocular surface of exogenously added fluorescein.
- Conjunctival biopsy: mean change in grades of squamous metaplasia of the conjunctiva.
We tabulated adverse effects (e.g. bacterial and viral infection and eye irritation) reported in the included studies for both the AS and control groups.
Quality of life measures
We planned to record health-related quality of life data presented by any validated measure (e.g. activities of daily vision scale) in the included studies.
We planned to documented cost analyses and other data on economic outcomes reported by the included studies.
Search methods for identification of studies
We searched the Cochrane Central Register of Controlled Trials (CENTRAL) 2013, Issue 3, part of The Cochrane Library. www.thecochranelibrary.com (accessed 15 April 2013), Ovid MEDLINE, Ovid MEDLINE In-Process and Other Non-Indexed Citations, Ovid MEDLINE Daily, Ovid OLDMEDLINE, (January 1950 to April 2013), EMBASE (January 1980 to April 2013), Latin American and Caribbean Literature on Health Sciences (LILACS) (January 1982 to April 2013), the metaRegister of Controlled Trials (mRCT) (www.controlled-trials.com), ClinicalTrials.gov (www.clinicaltrials.gov) and the WHO International Clinical Trials Registry Platform (ICTRP) (www.who.int/ictrp/search/en). We did not use any date or language restrictions in the electronic searches for trials. We last searched the electronic databases on 15 April 2013.
See: Appendices for details of search strategies for CENTRAL (Appendix 1), MEDLINE (Appendix 2), EMBASE (Appendix 3), LILACS (Appendix 4), mRCT (Appendix 5), ClinicalTrials.gov (Appendix 6) and the ICTRP (Appendix 7).
Searching other resources
We also searched the Science Citation Index-Expanded database (September 2013) and reference lists of included studies. We did not handsearch conference proceedings or journals.
Data collection and analysis
Selection of studies
Two review authors independently reviewed the titles and abstracts of all the reports identified from the electronic searches. We classified each study as 1) eligible for inclusion, 2) unsure, or 3) exclude. We obtained full-text copies of all potentially and definitely relevant articles. Two review authors assessed the full-text articles for final inclusion of studies in this review. For studies that we excluded at this stage, we documented reasons for exclusion (see Characteristics of excluded studies). We resolved any discrepancies through consensus.
Data extraction and management
Two review authors extracted the data independently using the data extraction form developed by the Cochrane Eyes and Vision Group for this review. We resolved discrepancies by discussion and contacted study authors for additional necessary data. All data were entered into Review Manager 5 (RevMan 2012) by one review author and confirmed by a second review author.
Assessment of risk of bias in included studies
Two review authors assessed risk of bias independently according to methods set out in Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a). Review authors were not masked to any trial details during the assessment. We considered the following risk of bias parameters for each of the included studies: sequence generation and allocation concealment (selection bias); masking (blinding) of participants and researchers during and after treatment as well as during outcome assessment (detection bias); completeness of follow-up for primary and secondary outcomes (attrition bias); and selective outcome reporting (reporting bias). We applied a judgment of 'low risk', 'unclear risk', or 'high risk' to each of the above parameters for each of the included studies.
For cross-over trials we considered additional methodological assessments of the risk of bias, including whether there was a wash-out period, the number lost to follow-up after each phase, and whether the data were reported for each phase or by treatment as described in Chapter 16 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b).
Measures of treatment effect
We did not conduct summary meta-analyses of the treatment effects in this review. If sufficient data are available in future updates we will calculate summary risk ratios (RRs) for dichotomous outcomes of interest (proportion of participants reporting improvement in dry eye related symptoms). We will summarize continuous data from objective ocular tests by calculating mean differences from baseline to follow-up between the treatment and control arms (ocular surface staining, Schirmer's test, and tear break-up time). For continuous scales of participant-reported outcomes, we will calculate standardized mean differences (SMDs) to account for the variation in measurement scales. We will dichotomize ordinal data to reflect varying degrees of symptom improvement ('some improvement') followed by sensitivity analyses using different cut points (Patrick 2011).
We will use the generic inverse variance method to summarize the treatment effects from studies that reported the computed measures of effect and variance estimates. We will not include quantitative data from cross-over trials which report only the first phase data, given the risk of bias for incomplete outcome reporting (Higgins 2011b).
Unit of analysis issues
The unit of analysis was the individual participants who were randomized to each treatment arm in two trials (Kojima 2005a; Urzua 2012). One trial used a paired-eye design in which each eye of the participant was evaluated and the eye was considered the unit of analysis. Another trial randomized participants to each intervention while the analyses included both eyes of each participant independently (Noda-Tsuruya 2006). We reported results using the unit of analysis reported by the studies.
Dealing with missing data
We contacted study authors of included trials for clarification or retrieval of missing primary and secondary outcome data. We did not conduct any imputations when study authors did not provide missing data and instead relied on data in the published reports. For future summary meta-analyses, when trial authors are unable to provide information on missing data, we plan to conduct the following sensitivity analyses: (a) assume all participants with missing data in the treated group had the worse outcome (if dichotomous); and (b) assume all participants with missing data in the treated group did not have the worse outcome.
Assessment of heterogeneity
We assessed clinical and methodological heterogeneity by examining the characteristics of study participants, treatment/control comparisons, and assessment of primary and secondary outcomes.
If future updates of this review include summary meta-analyses, we will examine consistency across studies with the I² test (Higgins 2003), with a value greater than 50% indicating substantial statistical heterogeneity. We will also inspect forest plots for the degree of overlap of the confidence intervals of the included studies. Little overlap is another indication of the presence of heterogeneity.
Assessment of reporting biases
We were not able to conduct summary meta-analyses and could not assess reporting bias through the inspection of funnel plots.
There were insufficient data to conduct a meta-analysis. A narrative summary of results was used in place of statistical summary analyses.
For future updates we will conduct a random-effects meta-analysis when there is significant clinical, methodological, and statistical homogeneity among included studies. When fewer than three studies are included in a meta-analysis, we will use a fixed-effect model. We will not combine studies in a meta-analysis when there is significant heterogeneity detected among included studies.
Subgroup analysis and investigation of heterogeneity
There were insufficient data to conduct a subgroup analysis in this review. If adequate data are present in future updates we will stratify by the underlying etiology of dry eye symptoms including tear deficiencies (Sjögren's syndrome), non–Sjögren’s syndrome–related dry eye, evaporative dry eye (blepharitis or meibomian gland dysfunction (MGD)), and complications of LASIK.
We did not conduct a sensitivity analysis in this review. For future updates we will investigate the impact of studies with lower methodological quality (i.e. high risk of bias for random sequence generation or incomplete outcome data for primary or secondary outcomes) and unpublished studies through sensitivity analyses.
Description of studies
Results of the search
We identified a total of 402 titles and abstracts from the electronic searches as of April 2013 (Figure 1). After removing the duplicates, we screened 360 titles and abstracts. We identified 30 reports of 29 studies as potentially relevant for this review. After full-text review of the 30 reports, we included three full-text reports from three trials (Noda-Tsuruya 2006; Tananuvat 2001; Urzua 2012) and one full-text report and conference abstract reporting findings from another trial (Kojima 2005a) (see Characteristics of included studies). We found no other eligible trials or additional reports of the four included trials after searching other sources described above.
|Figure 1. Results from searching for studies for inclusion in the review|
All study participants in the four trials included in this review had dry eye. The etiologies of dry eye were mainly post-LASIK, Sjögren's syndrome and non-Sjögren's syndrome. The number of participants in the studies ranged from 12 to 27 with the average age ranging between 30 and 60 years among three studies (Noda-Tsuruya 2006; Tananuvat 2001; Urzua 2012) and one with an overall age range from 50 to 75 years (Kojima 2005a).
Kojima 2005a included 20 participants with severe dry eye, 17 of whom had Sjögren's syndrome. None of the study participants had a history of ocular surgery or procedures, including punctal occlusion.
Noda-Tsuruya 2006 included 27 men who developed dry eye subsequent to LASIK and had not worn contact lenses before LASIK.
Tananuvat 2001 enrolled 13 study participants with severe dry eye including nine with a history of punctal occlusion. Five participants had Sjögren's syndrome, three had idiopathic dry eye, one had non-Hodgkin's lymphoma, one had graft-versus-host disease, one had Stevens-Johnson syndrome, and one had rheumatoid arthritis. One participant was excluded after enrollment due to an imbalance in the severity of dry eye between the two eyes.
Urzua 2012 enrolled 12 participants with severe non-Sjögren dry eye who had received prior treatment with artificial tears.
All four trials evaluated 20% AS with instructions given to participants to apply drops four, five, or six times daily. Similar instructions for the storage of AS study vials were given to participants in all four trials. In one trial, participants were given 5 ml bottles of 20% AS in unpreserved normal saline solution or bottles of saline solution mixed with dilute fluorescein solution, which served as placebo. Participants were instructed to use the eye drops six times per day for two months, and to refrigerate the eye drop bottle in use while the rest were frozen. One bottle of eye drops was to be used for one week and then replaced. They were also instructed to continue use of preservative-free artificial tears as needed (Tananuvat 2001).
In the second trial, participants entered a “wash-out” phase where they used only preservative-free saline eye drops six times a day for two weeks. Subsequently, participants in the AS group used only 20% AS in saline six times a day for two weeks, and participants in the artificial tear group used only preservative-free artificial tears six times a day for two weeks. Participants were instructed to keep the vials they were using in a refrigerator at 4°C and were instructed to store the other vials in a freezer (Kojima 2005a). The specific formulation of the artificial tears used by the control group was not reported.
In the third trial, after LASIK, all participants used low-dose steroids (0.1% fluorometholone, Flumetholon, Santen); antibiotics (Tarivid, Santen); and 0.3% hyaluronic acid (Hyalein, Santen) eye drops five times per day for one week (Noda-Tsuruya 2006). Subsequently, the AS group used eye drops made of 20% AS diluted in sterile saline five times a day from one week to six months postoperatively and the artificial tear group used preservative-free saline-based artificial tears (Soft Santear, Santen), five times a day from one week to six months postoperatively. Participants were instructed to keep the bottle they were using in a refrigerator at 4°C and were instructed to store the other bottles in a freezer (-20°C). Each bottle of 20% AS was used for two weeks and then replaced.
In the fourth trial, Urzua 2012, participants were given 14 identical flasks containing either 20% AS or artificial tears (Systane) and were instructed to use one flask four times a day for the first two weeks. After the first two weeks, all study participants used 0.9% NaCl for a one-week wash-out, and were then given another 14 flasks containing the second study intervention (either 20% AS or artificial tears), opposite of the intervention they received in the first two-week period.
Outcome Assessment Measures
Each of the four included trials used a different method to evaluate participant-reported symptom improvement at different follow-up times. In each method used, higher values represented more severe symptoms/discomfort in which a decrease in values from baseline would suggest improvement in symptoms. Only one study reported participant-report symptoms at one-month follow-up, the primary outcome for this review. In the other three trials, participant-reported symptoms were reported at additional follow-up periods between two weeks and six months follow-up. Tananuvat 2001 assessed study participants at baseline and on three follow-up visits at one week, one month, and two months. Symptoms of dry eye (discomfort, foreign-body sensation, dryness, and photophobia) were graded as grade 0, no symptoms; 1, mild; 2, moderate; and 3, severe. In Kojima 2005a, visual analog pain symptom scores were assessed at baseline and at two weeks. In Noda-Tsuruya 2006, a written questionnaire was used to assess dry eye symptoms; the participants graded “typical dry eye symptoms” 0, none; 1, mild; 2, moderate; 3, strong; and 4, very strong. In Urzua 2012, the Ocular Surface Disease Index (OSDI), recommended by the International Dry Eye Workshop (Ozcura 2007), was used to evaluate participant-reported improvement in dry eye symptoms.
Although all of the studies included TBUT, tear secretion (Schirmer's test) and fluorescein staining, the investigators of these studies did not all follow the same procedures with additional variation in the time points at which all outcomes were evaluated. We do not believe the variation in procedures used to evaluate the objective clinical tests would influence the ability to compare treatment effects across studies. In Kojima 2005a and Noda-Tsuruya 2006 TBUT was observed after instilling 2 μl of 1% Rose Bengal mixed with 1% fluorescein and saline into the cul-de-sac; in Tananuvat 2001, a fluorescein strip moistened with saline was placed into the lower cul-de-sac. In Urzua 2012, no additional description was given of how the investigators evaluated TBUT. Tananuvat 2001 and Noda-Tsuruya 2006 specify that the Schirmer test was done with anesthesia and without anesthesia in Kojima 2005a. Scoring of fluorescein staining of the ocular surface in Kojima 2005a and Noda-Tsuruya 2006 was carried out by dividing the cornea into upper, middle, and inferior compartments and grading each one on a scale of 0 to 3 points (maximum: 9 points). Tananuvat 2001 did not divide the cornea into thirds, and fluorescein staining of the cornea was graded from 0 to 3. Urzua 2012 used the OXFORD scale (six categories) to evaluate fluorescein staining (Bron 2003). Details of the procedures used to evaluate Rose Bengal staining were not reported by the three trials measuring this outcome (Kojima 2005a; Noda-Tsuruya 2006; Tananuvat 2001). Conjunctivel impression cytology and frequency of other topical lubricants were evaluated in only one trial (Tananuvat 2001).
We excluded 25 references after full-text review (see Characteristics of excluded studies). Two references were from conference abstracts (Harritshoj 2011; Jaksche 2005), and the remainder were from full-text publications. A majority of the excluded studies were non-randomized studies or reviews. We excluded three RCTs because the investigators did not compare AS to artificial tears or placebo (Jaksche 2005; Noble 2004; Yoon 2007).
Risk of bias in included studies
Figure 2 presents a summary of the risk of bias for the included studies. For two studies (Kojima 2005a; Noda-Tsuruya 2006) a majority of the risk of bias domains were unclear due to insufficient description in trial reports.
|Figure 2. Methodological quality summary: risk of bias review authors' judgements about each risk of bias item for each included study.|
Sequence generation and allocation concealment
The risk of bias domains for sequence generation and allocation concealment were judged to be at low risk of bias in one study (Urzua 2012) and at unclear risk for two included studies (Kojima 2005a; Noda-Tsuruya 2006). Although randomization of participants was specified in the later two trials, none of the published reports described the methods used to generate the allocation sequence or how they implemented the treatment allocation in sufficient detail. One study used block randomization with block sizes of two resulting in alternating treatment assignment which we judged to have a high risk of bias for sequence generation. As the investigators were unmasked and assignments in a block size of two could be known we judged this study have a high risk for allocation concealment (Tananuvat 2001).
Masking (performance bias and detection bias)
Masking of participants and study personnel
Masking of participants and study personnel to the allocated intervention was judged to be at low risk of bias in one study (Urzua 2012) and at unclear risk in two studies (Kojima 2005a; Noda-Tsuruya 2006). A full description of the measures used to achieve masking was not included in the published reports for Kojima 2005a and Noda-Tsuruya 2006 which specified a prospective randomized design without clearly reporting masking of participants or study personnel. The third trial was conducted as a single masked study with participants masked to their treatment assignments and investigators unmasked (Tananuvat 2001). We judged this trial to be at low risk of bias for masking participants and at high risk of bias for masking study personnel.
Participants allocated to the AS group had to undergo blood extraction as part of the serum production process. Specific instructions for the preservation and storage of AS were reported in all four studies (Kojima 2005a; Noda-Tsuruya 2006; Tananuvat 2001; Urzua 2012). It is not clear at what point in the randomization process participants were subjected to serum collection procedures and whether the same storage instructions were provided to all participants regardless of their treatment assignment. One study (Urzua 2012) implemented a cross-over study design which maintained participant masking whereby all participants underwent venous blood draw for preparation of their AS. Additional measures taken in Urzua 2012, including the use of opaque flasks and instructions to keep all study medication frozen at -20°C helped ensure participant masking.
Masking of outcome assessors
Outcome assessments were considered in two main categories: 1) assessment of participant-reported symptoms and 2) assessment of objective clinical examination. For two studies, the masking of outcome assessors for participant-reported symptoms was judged to be unclear (Kojima 2005a; Noda-Tsuruya 2006). Neither study provided a full description of how the participant-reported outcomes were recorded and whether the study personnel collecting this information were aware of the participant’s treatment assignment. Participants were asked to complete either a written questionnaire or an analog pain scale in two studies (Kojima 2005a; Noda-Tsuruya 2006). In one study, participants were asked to report symptoms at each visit, and this information was then recorded by an unmasked study investigator and was judged to be at high risk of bias (Tananuvat 2001). Masking of outcome assessors for the objective clinical examination was judged as unclear for one study (Noda-Tsuruya 2006) and one study was judged as being at low risk of bias (Kojima 2005a). Study investigators were unmasked in one study which was determined to be at a high risk of bias for the objective clinical outcome assessment (Tananuvat 2001). Although these were objective clinical tests, there is potential detection bias if investigators conducting the test and interpreting the results were aware of the participant's treatment assignment. Urzua 2012 was the only study found to be at low risk of bias for both participant-reported outcomes and objective clinical tests.
Incomplete outcome data
The domain for incomplete outcome data was judged to be at low risk of bias for two studies (Tananuvat 2001; Urzua 2012). There was no loss to follow-up or missing data reported as confirmed by a review of the number analyzed after initial inclusion/exclusion in the results section. One study (Noda-Tsuruya 2006) reported the number of eyes for each outcome at all time points across both treatment arms, but the investigators did not provide reasons for missing outcome data; the number of analyzed eyes was variable throughout the intervention and resulted in a judgment of unclear risk of bias. Two eyes were excluded from the analyses in the full-text report from one trial (Kojima 2005a) compared to the conference abstract for the same trial, without an explanation for the discrepancy.
We found two studies to be at low risk of reporting bias (Kojima 2005a; Urzua 2012); the other two studies were judged to have an unclear risk of reporting bias (Noda-Tsuruya 2006; Tananuvat 2001); the investigators reported all outcomes at all time points as described in the methods, although reported information was insufficient to extract usable data for quantitative summary analyses. For one study we were able to confirm the prespecified outcomes described in the ClinicalTrials.gov record with the corresponding publication (Urzua 2012), but did not have access to study protocols or other related materials for the other three studies, and were unable to confirm the reported outcomes with the intended outcomes for each study.
Other potential sources of bias
We were unable to fully assess other potential sources of bias for two studies that were judged to have an unclear risk of bias. In one study (Tananuvat 2001), participants in both groups were able to use artificial tears lubricants. The estimated treatment effect of AS may be influenced if the additional lubricants had a perceived therapeutic effect on the outcomes of interest and were used in different frequencies in each group. In another study (Noda-Tsuruya 2006), there was a discrepancy between the unit of randomization (individual) and unit of analysis (eyes), which can lead to biased treatment effects by not considering the within participant correlation. For one study we found sufficient information (appropriate study design, proper ethical conduct, no involvement from industry) to establish a low risk for other potential bias (Kojima 2005a). On review of the predefined inclusion criteria, we identified a discrepancy with the inclusion criteria listed in the published report for one study which led us to judge this as an unclear risk of bias (Urzua 2012).
Effects of interventions
Included studies could not be combined in summary meta-analyses due to heterogeneity in the time points at which primary and secondary outcomes were reported and insufficient reporting of descriptive statistics (means and standard deviations) necessary for computing treatment effect estimates. One study implemented a cross-over design but did not report the necessary summary statistics from a paired analysis (i.e. mean difference from paired t-test and corresponding confidence interval or P value) to account for the participant-level differences in AS and artificial tears (Urzua 2012). We can therefore provide only a narrative description of the reported findings.
None of the included trials reported results for the primary outcome of this review, the change from baseline in participant-reported symptoms at one month follow-up.
At one month follow-up, the mean composite symptom score was 5.36 and 6.45 for both the 20% AS and control group (saline solution with diluted fluorescein) respectively, and 5.3 and 5.9 at the two-month follow-up visit for the 20% AS and the control group respectively (Tananuvat 2001). Without the reported number of participants and SD estimates for each treatment group we could not generate an overall variance estimate (95% CI) and corresponding P value for the difference between 20% AS and the control group for either follow-up interval ( Analysis 2.1). The authors reported the mean symptom scores were not statistically different (P > 0.05) between the 20% AS and control groups over a two-month treatment period.
Among 10 participants in both the 20% AS group and the artificial tear control group, the mean change and SD from baseline at two weeks follow-up measured with the visual analog scale was -19.2 ± 8.8 and -7.2 ± 9.8 and respectively, resulting in a difference in the mean change from baseline of -12.00 (95% confidence interval (CI) -20.16 to -3.84 ( Analysis 1.1; Kojima 2005a). This difference suggests there was a greater decrease in the pain/dry eye symptoms in the 20% AS group compared to the artificial tear group after two weeks. The mean and SD in the OSDI scale in the 20% AS and artificial tear groups was 59 ± 10 and 51 ± 7, respectively, among 12 participants at baseline and 30 ∓ 8 and 41 ± 8, respectively, at two weeks (Urzua 2012). From these data, the trial investigators reported there was a 51% decrease in the OSDI scale in the 20% AS group and a 22% decrease in the artificial tear group.
Descriptive statistics (mean and SD) were not reported for 27 post-LASIK participants (54 eyes) as measured by a five-point questionnaire (Noda-Tsuruya 2006). However, in a narrative description, the trial investigators reported that there was no statistically significant difference (P > 0.05) in the participant-reported symptoms between the 12 participants (24 eyes) in the 20% AS group and the 15 participants (30 eyes) in the artificial tear group before and after LASIK surgery through six months of follow-up.
Ocular surface staining
Rose Bengal staining
The mean changes and SDs from baseline in Rose Bengal at two weeks follow-up for the 20% AS and artificial tear groups were -2.3 ± 0.8 and -0.1 ± 0.3 respectively, resulting in a difference in the mean change from baseline of -2.20 (95% CI -2.73 to -1.67; P < 0.00001) ( Analysis 1.2) across 20 participants (10 in each treatment group) (Kojima 2005a). The difference in the decrease in Rose Bengal scores suggest the 20% AS group had greater overall improvement compared to the artifical tear group after two weeks.
The means and SDs were 0.3 ± 0.7 in 20 eyes in the 20% AS group and 0.9 ± 0.8 in 15 eyes in the artificial tear group, resulting in a mean difference in Rose Bengal staining score of -0.60 (95% CI -1.11 to -0.09; P = 0.02) one month after LASIK ( Analysis 1.3), and 0.1 ± 0.3 (16 eyes) and 0.8 ± 1.1 (11 eyes), respectively, three months after LASIK resulting in a mean difference of -0.70 (95% CI -1.37 to -0.03; P = 0.04) ( Analysis 1.3; Noda-Tsuruya 2006).
The reported means for Rose Bengal in the 20% AS group at one week, one month and two months of follow-up were 3, 4.22 and 3.7, and 3.67, 4.22 and 3.8 in the control group respectively (Tananuvat 2001). Again, the number of participants and SDs were not reported, although the trial investigators indicated a non-significant difference (P > 0.05) between 20% AS and artificial tears across all time points ( Analysis 2.2).
One study reported fluorescein staining greater than 1 as abnormal (Kojima 2005a). In 10 participants, the mean change and SD in fluorescein staining from baseline to two weeks follow-up was -1.1 ± 0.7 in the 20% AS group and -0.2 ± 0.6 in the artificial tear group, resulting in a difference in the mean change from baseline of -0.90 (95% CI -1.47 to -0.33; P = 0.002) ( Analysis 1.4; Kojima 2005a). Noda-Tsuruya 2006 reported the mean and SD among 20 eyes at one month follow-up for the 20% AS group (0.5 ∓ 0.7), but the mean from the 23 eyes in the control group was not reported. Among 12 participants, the reported means at one week, one month and three months follow-up were 1.6, 1.55 and 1.33 in the 20% AS group and 1.7, 1.55 and 1.42 for the control group respectively ( Analysis 2.3; Tananuvat 2001). The mean OXFORD scale for fluorescein staining decreased from baseline to two weeks from three to two in the 20% AS group, and from four to three in the artificial tear group (Urzua 2012). The trial investigators reported this was a non-significant difference (P > 0.05).
Aqueous tear production: Schirmer’s test
Schirmer’s test was performed without anesthesia in one of the included studies (Kojima 2005a) and with anesthesia in two of the included studies (Noda-Tsuruya 2006; Tananuvat 2001). At two weeks follow-up, the mean and SD for the 20% AS group was 3.3 ± 2.6 mm compared to 3.7 ± 3.1 mm in the artificial tear group, resulting in a mean difference of -0.40 (95% CI -2.91 to 2.11 mm; P = 0.75) ( Analysis 1.5; Kojima 2005a). Tananuvat 2001 reported means of 2.83 mm in the 20% AS group and 3.25 mm in the control group after two months of follow-up, compared to 0.92 mm and 1.83 mm for both groups at baseline respectively ( Analysis 2.4). Discriptive statistics (mean and SD) were not reported by Noda-Tsuruya 2006 for the 20 eyes in the 20% AS group or the 15 eyes in the control group evaluated for this outcome at one month. The trial investigators concluded that there was not a significant difference between the two groups.
Tear film stability: tear break-up time (TBUT)
Kojima 2005a reported a mean change and SD from baseline to two weeks follow up of 2.1 ± 1.1 seconds in the 20% AS group and 0.1 ± 1.2 seconds in the artificial tear group, resulting in a mean difference of 2.00 (95% CI 0.99 to 3.01) seconds (P = 0.0001) between 10 participants in each treatment group ( Analysis 1.6). Among the 12 participants in Urzua 2012, the baseline mean TBUT was 4 ± 1.9 seconds in the 20% AS group and 3 ± 2.2 seconds in the artificial tear group. After two weeks, the mean was 6 ± 1.2 seconds in the 20% AS group and 4 ± 2.3 seconds in the artificial tear group (P > 0.05). Tananuvat 2001 described means (in seconds) of 0.8 at one week, 0.55 at one month and 0.83 at two months for the 20% AS group and 0.7 at one week, 0.64 at one month and 1.17 at two months for the control group ( Analysis 2.5). The six-month mean TBUT for the 20% AS groups was 6.3 ± 2.6 seconds among eight postoperative LASIK eyes in the 20% AS group and 3.8 ± 1.9 seconds in 10 control eyes, leading to a mean difference of 2.50 (95% CI 0.35 to 4.65) seconds (P = 0.02) ( Analysis 1.7; Noda-Tsuruya 2006).
Only one study reported results from impression cytology at baseline and at two months follow-up according to conjunctival differentiation separated into six stages scored from 0 to 6 (Tananuvat 2001). At baseline, the mean score was 2.67 in the 20% AS group and 2.42 in the artificial tear group followed by means of 1.57 in the 20% AS group and 2.17 in the control group after two months of follow-up. The trial investigators reported that the mean difference at two months follow-up was non-significant (P > 0.05).
Complications and infections
Tananuvat 2001 reported two participants with signs of conjunctivitis with negative culture; in both cases symptoms resolved later with proper treatment. It was not stated whether the eye assigned to the AS or control group, or both eyes showed signs of conjunctivitis. Microbiologic culture of serum stored at -20°C for up to two months showed no growth. All returned used serum bottles underwent culture, and only one sample exhibited mixed organisms, including yeast. No infectious conjunctivitis or other adverse reaction was detected. Adverse events were not reported by the investigators of the remaining trials (Kojima 2005a; Noda-Tsuruya 2006; Urzua 2012).
None of the included trials evaluated tear osmolarity, corneal topography, fluorescein clearance, or conjunctival biopsy, specified as secondary outcomes for this review. Quality of life and cost or economic analyses were not reported by any of the included studies as well.
The use of autologous serum eye drops (AS) to treat people with dry eye has been described in a number of studies (Fox 1984; Kojima 2005a; Noda-Tsuruya 2006; Tananuvat 2001; Tsubota 1996; Tsubota 1999; Tsubota 2000). Our aim in performing this systematic review was to analyze the highest quality evidence from randomized controlled trials (RCTs) to determine the efficacy and safety of AS in treating people with dry eye. However, the majority of the published literature is limited to retrospective case reports or non-randomized case series.
Summary of main results
We identified four RCTs that investigated the effects of AS compared to artificial tears in participants with a variety of dry eye etiologies (Kojima 2005a; Noda-Tsuruya 2006; Tananuvat 2001; Urzua 2012). Kojima 2005a evaluated the effectiveness of 20% AS after a two-week treatment interval (six times a day) in participants with severe Sjögren's and non-Sjögren's syndrome dry eye. Urzua 2012 used a cross-over design to compare two week treatment intervals with 20% AS and artificial tears in 12 adult participants with severe non-Sjögren's syndrome dry eye. Tananuvat 2001 investigated efficacy of 20% AS in 12 participants with bilateral severe dry eye over a two-month treatment interval (six times daily). Noda-Tsuruya 2006 assessed the efficacy of 20% AS (five times daily) for post-LASIK dry eye from one week to six months.
Although precise measurement of symptoms is an important part of dry eye diagnosis, there is no universally accepted standardized method for recording participant-reported symptoms; it is commonly observed that participant-reported symptoms do not correlate with objective clinical tests (Alfonso 1999; Lin 2003; Schein 1997; Viso 2012). In this review, each study applied different methods to measure participant-reported symptoms. After taking into consideration the wide array of subjective questionnaires and scales used to measure participant symptoms and differences in the length of follow-up, improvement in participant-reported symptoms was not consistently observed in the four trials comparing AS to artificial tears. This might be due to the variety in the type and severity of dry eye among the participants in the studies included in this review.
Based on the reported data from the included studies, 20% AS were not associated with a significant improvement in aqueous tear production as measured by Schirmer’s test or improvement in the condition of the ocular surface as measured by fluorescein or Rose Bengal staining compared to preservative-free artificial tears. Tananuvat 2001 further found that 20% AS did not significantly change impression cytology among participants with severe bilateral dry eye. Regarding tear film stability as measured by tear break-up time (TBUT), only Kojima 2005a showed a clinically meaningful difference between 20% AS and artificial tears for Sjögren's and non-Sjögren's syndrome dry eye participants after two weeks of treatment.
Three of the included studies did not report outcomes for adverse events or complications. One study (Tananuvat 2001) reported conjunctivitis in two participants with cultures showing no growth followed by resolution of symptoms. All used AS containers returned by study participants were cultured, with one sample showing mixed organisms, including yeast, but no infectious conjunctivitis or adverse reaction was detected in the study participant.
Overall completeness and applicability of evidence
A major difficulty in summarizing results from the included studies was the heterogeneity in the participant populations, interventions and comparisons, as well as variations in the procedures for preparing AS. Summary meta-analyses could not be conducted due to additional differences in follow-up intervals as well as incomplete descriptive statistics in the reported treatment outcomes. Thus, we were able to draw conclusions based on qualitative assessment of the trial reports.
The etiologies of dry eye described for included participants may not be representative of all people with dry eye who potentially may benefit from AS. Previous punctal occlusion was reported as an exclusion criterion in one study (Kojima 2005a), while people with previous punctal occlusion were eligible for another (Tananuvat 2001). One trial included participants with post-LASIK dry eye (Noda-Tsuruya 2006). Two trials enrolled participants with both severe Sjögren's and non-Sjögren's syndrome dry eye (Kojima 2005a; Tananuvat 2001), while participants with severe non-Sjögren's syndrome dry eye only were enrolled in another (Urzua 2012).
Interventions and comparisons
It is worth noting that participants of the intra-individual study (Tananuvat 2001) in which participants used AS in one eye and placebo in the fellow eye were instructed to use non-hyaluronan and unpreserved saline-based artificial tears as needed. Prior punctal occlusion was also reported in 75% of participants at the beginning of the study, adding further to the heterogeneity among included studies (Tananuvat 2001).
Preparation and storage of AS
Currently there are neither regulatory guidelines nor standard protocols for the manufacturing of AS for dry eye. The critical steps in the production of AS, such as clotting time, centrifugation, and dilution can influence the biochemical properties of AS and may lead to variable efficacy and treatment outcomes. In this review, only one study reported a clotting time of two hours following venipuncture (Urzua 2012) while clotting time was not reported in the other three studies. Geerling 2004 in Germany proposed two hours of clotting time at room temperature followed by optimal centrifugation of whole blood at 3000 × g for 15 minutes. Variation in the centrifugation speed and time were reported across included trials ranging from 1500 revolutions per minute (rpm) for five minutes (Kojima 2005a), 2200 rpm for 20 minutes (Noda-Tsuruya 2006), 4200 rpm for 15 minutes (Tananuvat 2001), and 3500 rpm for five minutes (Urzua 2012). It has been demonstrated that higher concentrations of EGF and lower concentrations TGF-b are obtained at higher centrifugation speed (Liu 2005; Pancholi 1998; Phasukkijwatana 2011). Although none of the trials measured the concentration of the biologically active components within AS, it is possible that the concentration of epithelial growth factor (EGF), transforming growth factor-β (TGF-β) or other biologic factors might be different across the included studies due to variation in the rpm used in centrifugation. Interestingly, all of the four included trials compared 20% AS to non-hyaluronan and unpreserved saline-based artificial tears in this review. Although results from in vitro studies show the greatest cell proliferation with serum concentrations ranging from 12.5% to 25% (Geerling 2004; Liu 2005), 20% AS was the only concentration evaluated in the four trials included in this review.
In addition, the instructions given to study participants for storing AS were similar across included studies. Specifically, they were instructed to keep vials containing AS in the freezer (-20°C) for up to three months and in a refrigerator at 4° C for two weeks after thawing. The AS storage instructions given to study participants have been shown to be effective in preventing contamination and deterioration of biological growth factors (Geerling 2004). Kojima 2005a reported an additional precaution to protect serum vials from ultraviolet light because vitamin A is easily degraded by light. A study from Thailand (Phasukkijwatana 2011) demonstrated that the stability of biologically active components within AS could be maintained for up to six months when stored at -20°C. However, the U.S. Food and Drug Administration has not approved a standard procedure for preparing AS.
Quality of the evidence
Only four small RCTs in single-center settings have been identified; they included 72 total participants with severe Sjögren's-related dry eye, non-Sjögren's dry eye and post-LASIK dry eye. The small number of participants is insufficient to detect or rule out meaningful beneficial or harmful effects of AS or to provide precise estimates of individual outcomes.
Two studies were found to have an unclear risk of bias for masking of participants. Given the primary outcome (i.e. change in participant-reported symptoms), the results of individual trials could have been influenced if participants were aware of their treatment assignment. However, complete masking may not be feasible given the necessary venipuncture involved in AS production. Additional variation in the instructions reported for the proper storage of AS compared to artificial tears may also have led to participants knowing their treatment assignment.
Potential biases in the review process
We employed a comprehensive search strategy to identify potentially eligible trials in order to minimize selection bias. Throughout the review process, two review authors assessed all potentially eligible studies and completed data extraction independently to minimize errors. Although we sought unpublished data from investigators of all the included trials to supplement the data provided in the published reports, we were unable to conduct quantitative synthesis for any of the outcomes specified for this review. Based on the limited number of included trials we could not evaluate potential publication bias through examination of funnel plots.
Agreements and disagreements with other studies or reviews
In the 2011 Prefered Practice Patterns, the American Academy of Ophthalmology (AAO) suggested autologous serum drops improve ocular symptoms and conjunctival and corneal staining in severe dry eye (AAO 2011). The AAO's conclusions were described as level "A III" and did not incorporate the findings or conclusions from any of the four RCTs included in our review. Another evidence-based review (Akpek 2011) found II B evidence for serum eye drops in Sjögren's syndrome dry eye which reflect an absence of reliable evidence to support treatment decisions. None of the four trials included in our review were discussed in the review by Akpek et al. (Akpek 2011) which was focused on individuals with Sjögren's syndrome dry eye only and evaluated two studies excluded from our review (Noble 2004; Yoon 2007).
Implications for practice
Based on the current evidence, 20% AS may provide some benefit in improving participant-reported symptoms in the short term (two weeks), but improvement was not observed through longer periods of follow-up. No effect was seen based on objective clinical measures of the ocular surface.
AS preparation, manufacturing and storage require a well-established, specialized service with strict aseptic processing. Procedures for AS production (clotting time, centrifugation and concentration), including the proper solute for making the AS should be optimized for the clinical application of AS in people with dry eye. In addition, all applicable legislative restrictions should be carefully considered and well documented, and informed consent should be obtained from each participant.
Implications for research
Well-planned, large scale, high-quality randomized controlled trials are needed, stratified by age and severity of dry eye, comparing AS to artificial tears (or other treatments), as well as evaluating additional concentrations of AS. These studies must have a random sequence generation protocol as well as appropriate concealment of the treatment assignments before allocation. Future studies should make attempts to ensure both participants and study investigators (clinical staff and outcome assessors) are masked to the treatment assignments in order to limit potential bias in participant-reported outcomes. We recommend that randomization in such trials be stratified by participant's age and the severity of dry eye-related symptoms. Any future studies should utilize standardized and validated scoring systems of dry eye clinical severity and symptom questionnaires. Objective biomarkers, which have been reported as a parallel index of the dry eye severity scale, such as tear osmolarity, tear cytokines and HLA-DR expression by ocular surface cells (Lemp 2011; Tomlinson 2006; Versura 2012), should be applied as outcomes in conjunction with participant symptoms. Analyses should include both short-term (two to four weeks) and long-term (six to 12 months) outcomes. Data on adverse outcomes, including complications, infection, and tolerance of AS should be documented in future trials.
We acknowledge Drs. Takashi Kojima and Napaporn Tananuvat for providing additional study data. We acknowledge Drs. Yassin Daoud and Manuel B. Datiles, co-authors of the published protocol, for providing guidance in the design of the review. We also acknowledge Bradley B. Brimhall, Richard S. Davidson, Vikram D. Durairaj, Xinggang Liu, and Swaroop Vedula for providing comments and methodological assistance in the development of the review protocol. We acknowledge Michael Lemp, Alan Sugar, Barbara Hawkins and the Cochrane Eyes and Vision Group (CEVG) editorial base for their comments on the protocol and review and for creating the electronic search strategies.
Richard Wormald (Co-ordinating Editor for CEVG) acknowledges financial support for his CEVG research sessions from the Department of Health through the award made by the National Institute for Health Research to Moorfields Eye Hospital NHS Foundation Trust and UCL Institute of Ophthalmology for a Specialist Biomedical Research Centre for Ophthalmology.
The views expressed in this publication are those of the authors and not necessarily those of the NIHR, NHS or the Department of Health.
Data and analyses
- Top of page
- Authors' conclusions
- Data and analyses
- What's new
- Contributions of authors
- Declarations of interest
- Sources of support
- Index terms
Appendix 1. CENTRAL search strategy
#1 MeSH descriptor: [Dry Eye Syndromes] explode all trees
#2 dry near eye*
#3 ocular near dry*
#4 MeSH descriptor: [Tears] explode all trees
#5t ear* near/2 film*
#8 sjogren* near syndrome
#9 steven* johnson syndrome*
#10 MeSH descriptor: [Pemphigoid, Benign Mucous Membrane] explode all trees
#11 cicatricial pemphgoid*
#13 MeSH descriptor: [Meibomian Glands] explode all trees
#16 #1 or #2 or #3 or #4 or #5 or #6 or #7 or #8 or #9 or #10 or #11 or #12 or #13 or #14 or #15
#17 MeSH descriptor: [Serum] explode all trees
#18 autologous near/2 serum*
#19 #17 or #18
#20 #16 and #19
Appendix 2. MEDLINE (OvidSP) search strategy
1. randomized controlled trial.pt.
2. (randomized or randomised).ab,ti.
9. exp animals/
10. exp humans/
11. 9 not (9 and 10)
12. 8 not 11
13. exp dry eye syndromes/
14. (dry adj2 eye$).tw.
15. (ocular adj2 dry$).tw.
16. exp tears/
17. (tear adj2 film$).tw.
20. Sjogren$ syndrome.tw.
21. Stevens Johnson syndrome/
22. Steven$ Johnson syndrome$.tw.
23. Pemphigoid, Benign Mucous Membrane/
24. cicatricial pemphigoid$.tw.
26. meibomian glands/
30. exp serum/
31. (autologous adj2 serum$).tw.
33. 29 and 32
34. 12 and 33
The search filter for trials at the beginning of the MEDLINE strategy is from the published paper by Glanville (Glanville 2006).
Appendix 3. EMBASE (OvidSP) search strategy
1. exp randomized controlled trial/
2. exp randomization/
3. exp double blind procedure/
4. exp single blind procedure/
7. (animal or animal experiment).sh.
9. 7 and 8
10. 7 not 9
11. 6 not 10
12. exp clinical trial/
13. (clin$ adj3 trial$).tw.
14. ((singl$ or doubl$ or trebl$ or tripl$) adj3 (blind$ or mask$)).tw.
15. exp placebo/
18. exp experimental design/
19. exp crossover procedure/
20. exp control group/
21. exp latin square design/
23. 22 not 10
24. 23 not 11
25. exp comparative study/
26. exp evaluation/
27. exp prospective study/
28. (control$ or prospectiv$ or volunteer$).tw.
30. 29 not 10
31. 30 not (11 or 23)
32. 11 or 24 or 31
33. dry eye/
34. (dry adj2 eye$).tw.
35. (ocular adj2 dry$).tw.
36. (tear adj2 film$).tw.
39. keratoconjunctivitis sicca/
41. Sjogren syndrome/
42. Sjogren$ syndrome.tw.
43. Stevens Johnson syndrome/
44. Steven$ Johnson syndrome$.tw.
45. mucous membrane pemphigoid/
46. cicatricial pemphigoid$.tw.
48. meibomian gland/
50. lacrimal apparatus/
51. lacrimal fluid/
54. exp serum/
55. (autologous adj2 serum$).tw.
57. 53 and 56
58. 32 and 57
Appendix 4. LILACS search strategy
dry eye and autologous
Appendix 5. metaRegister of Controlled Trials search strategy
dry eye and autologous
Appendix 6. ClinicalTrials.gov search strategy
Dry Eye AND Autologous
Appendix 7. ICTRP search strategy
dry eye OR dry eyes = Condition AND Autologous = Intervention
Last assessed as up-to-date: 15 April 2013.
Contributions of authors
Conceiving the review: AA
Designing the review: AA, QP
Coordinating the review: MM
Data collection for the review
- Designing search strategies: AA, CEVG Trials Search Co-ordinator
- Undertaking searches: CEVG Trials Search Co-ordinator
- Screening search results: QP, AA, AZ, MM, TH, LT
- Organizing retrieval of papers: MM
- Screening retrieved papers against inclusion criteria: QP, AA, AZ, MM, TH, LT
- Appraising quality of papers: QP, AA, AZ, MM
- Extracting data from papers: QP, AA, AZ, MM
- Writing to authors of papers for additional information: AZ, MM
- Providing additional data about papers: MM
- Obtaining and screening data on unpublished studies: AZ, MM
Data management for the review
- Entering data into RevMan: MM
- Checking data once entered into RevMan: QP
Interpretation of data
- Providing a methodological perspective: MM
- Providing a clinical perspective: QP, AA, AZ, YD, WS,TH, EKA
- Providing a policy perspective: QP, AA, YD, WS, EKA
- Providing a consumer perspective:
Writing the review: QP, AA, AZ, MM
Performing previous work that was the foundation of the current study: AA, EKA
Guarantor of the review: MM
Declarations of interest
No authors have conflicts of interest to report.
Sources of support
- This work was supported in part by the Raymond Kwok Family Research Fund, USA.
- Dr. Esen K. Akpek was supported in part by the Jerome L. Greene Sjogren's Syndrome Discovery Fund, USA.
- National Eye Institute, National Institutes of Health, USA.Michael Marrone was supported by the Cochrane Eyes and Vision Group US Project, which is funded by Grant 1 U01 EY020522-01
Medical Subject Headings (MeSH)
MeSH check words
* Indicates the major publication for the study