Novel Molecular Diagnostic System of Limbal Stem Cell Deficiency Based on MUC5AC Transcript Detection in Corneal Epithelium by PCR-Reverse Dot Blot

Abstract

Purpose.: To evaluate the sensitivity and specificity of a PCR-strip system based on reverse dot blot for detection of MUC5AC mRNA in corneal epithelium samples from patients with limbal stem cell deficiency (LSCD), and to determine the correlation with clinical diagnosis.

Methods.: We obtained 87 corneal impression cytology (IC) samples from 55 subjects (37 patients clinically diagnosed with LSCD and 18 control subjects). Total RNA was extracted from each IC sample and retrotranscribed to cDNA. MUC5AC mRNA transcript was amplified by a customized RT-PCR assay and detected in PCR strips based on reverse dot blot hybridization and in agarose gels. Conjunctival IC samples were used as positive controls.

Results.: Forty-four of 45 corneal IC samples obtained from patients clinically diagnosed with LSCD were positive for MUC5AC, whereas 34 of 42 corneal ICs from healthy subjects were negative for MUC5AC. Four healthy corneas were found MUC5AC positive, and four rendered inconclusive results. A correlation of 91.4% (P < 0.001) between molecular testing and clinical diagnosis was found. The overall sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of the PCR-strip system were 98%, 89%, 92%, and 97%, respectively.

Conclusions.: Corneal epithelium MUC5AC transcript detection by reverse dot blot PCR-strip is a highly sensitive technique for LSCD diagnosis. The test results strongly correlate with clinical diagnosis of characterized LSCD cases. The PCR-strip system may be used for early detection, and for the detection of mild cases of LSCD, and constitutes an objective clinical tool for the monitoring of treatments and surgical decisions.

Introduction

Limbal stem cell deficiency (LSCD) is a pathological eye condition in which the stem cells located at the limbus malfunction or are either partially or totally depleted, resulting in various degrees of stem cell deficiency and abnormalities at the corneal surface. Limbal stem cells act as a “barrier” to conjunctival epithelial cells preventing them from migrating onto the corneal surface. ^1,2 LSCD can have many causes and its severity can range from mild, as seen in contact lens overwear, to severe, as in chemical burns and ocular cicatricial pemphigoid (OCP). Causes of LSCD mainly include aniridia, ectodermal dysplasia, toxicity from topical medications, chemical or thermal injury, radiation, Stevens-Johnson Syndrome (SJS), OCP, cryotherapy, multiple surgery, and contact lens wear. ³  

Epithelial cells (ECs) of the ocular surface are characterized by their expression of the keratin antigens of intermediate filaments, which are highly tissue-specific molecular markers. The expression of K3/12 is currently thought to be a hallmark of corneal ECs, and has been used for LSCD diagnosis. ^4–10 However, the corneal specificity of K3 has recently been brought into question, with the consequence that the expression patterns of K3/12 in the ocular surface epithelium are not as clear-cut as believed. ^11–18 The diagnosis of LSCD has long relied on the detection of goblet cells in the IC of the corneal epithelium, either by periodic acid–Schiff staining ^12,19 or by more specific techniques, such as mucin 5AC staining. ^20,21 Since the MUC5AC transcript is considered to be a molecular marker for goblet cells, its detection in the corneal epithelium is indicative of LSCD. 

We have recently reported the molecular detection of the MUC5AC transcript in the cornea as a tool for diagnosing LSCD. ²² In that study, we evaluated the efficiency and reproducibility of the molecular technique in comparison with IC followed by PAS-hematoxylin staining. Having demonstrated the advantages of the RT-PCR method for LSCD diagnosis, a system for MUC5AC transcript detection in corneal epithelium was then developed to allow the diagnostic procedure to be performed in situ in hospital laboratories. This novel method is based on reverse dot-blot technology, and detection of the amplified fragment used for diagnosis is performed by visualizing a colorimetric band on a membrane. This type of PCR-strip–based diagnostic system is widely employed for the clinical diagnosis of several other several diseases, such as celiac disease, ²³ human papillomavirus, ²⁴ oligospermia, ²⁵ hemochromatosis, ²⁶ among others. The PCR-strip–based diagnostic system is easy to perform, does not require highly experienced personnel, and is routinely used in clinical diagnostic laboratories worldwide. 

The main objective of this study was to determine the sensitivity and specificity of the PCR-strip system for LSCD diagnosis by MUC5AC transcript amplification in the corneal epithelium and to correlate these results with clinical diagnosis. 

Methods

Patients

A transversal study was carried out in five hospitals: Cruces Hospital (Baracaldo, Vizcaya), Hospital Clínico (Barcelona), Instituto Fernández Vega (Oviedo), La Paz Hospital (Madrid), and Clínico San Carlos Hospital (Madrid). The work was conducted in two consecutive studies, analyzing a total of 87 eyes from 55 subjects (Table 1), including 37 patients clinically diagnosed with LSCD by means of slit-lamp examination (15 men, 22 women, with a mean age of 46.6 ± 13.3 years) and 18 healthy volunteers (including 7 men, 11 women, with a mean age of 32.1 ± 6.3 years). Patients were clinically diagnosed with LSCD due to a variety of etiologies including chemical burn, multiple glaucoma surgery, OCP, aniridia, Stevens-Johnson Syndrome, contact lens wear, ocular surface melanoma surgery and others (Table 2). IC samples were obtained after patients had signed an informed consent form in accordance with the principles of the Declaration of Helsinki on Biomedical Research Involving Human Subjects. 

Table 1

 

Descriptive Summary of the Samples Employed in the Study

Group	Subjects		Corneal IC Samples		Conjunctival IC Samples
	LSCD	Healthy	LSCD	Healthy
Study 1	15	7	19	15*	2
Study 2	22	11	26	27†	16
∑ =	37	18	45	42	18
Total	55		87		18

*  Fourteen samples from seven healthy patients (both eyes), and one from the healthy eye of a patient diagnosed with LSCD in the contralateral eye.

†  Twenty-two corneal IC samples from 11 healthy volunteers (both eyes) and five from the contralateral healthy eye of patients diagnosed with LSCD.

Table 2

 

Causes of LSCD

Condition/Cause of LSCD	Individuals, n
Chemical burn	9
Pemphigoid	2
Aniridia	1
Stevens-Johnson Syndrome	1
Contact lens wear	1
Ocular surface melanoma/carcinoma	2
Glaucoma surgery	1
Severe infections of ocular surface	3
Severe inflammations (of different origin)	3
Transplant failure due to herpes	1
Idiopathic	5
Others	8
Total	37

Impression Cytology Sample Collection

A representative illustration of impression cytology (IC) sample collection and sample processing is shown in Figure 1. IC samples were obtained using 8-mm diameter cellulose acetate discs (HAWP304; Millipore, Bedford, MA). Samples were always obtained in the following order: first from the cornea and then from the conjunctiva (when sampled), after having applied topical anesthesia to the eye surface, using a mixture of oxybuprocaine hydrochloride, tetracaine hydrochloride, and chlorobutanol (Colircusi double anesthetic; Alcon Cusí, Barcelona, Spain). The presence of goblet cells in the conjunctival samples was used as a positive control when available. Additionally, the specificity of MUC5AC transcript detection was verified in corneal samples from healthy volunteers (MUC5AC negative controls). 

Figure 1

 

Schematic representation of sample collection and processing. Corneal IC samples were collected using sterile tweezers without touching the conjunctiva. The membrane carrying the IC sample was immediately transferred to a RNase-free tube and store at 4°C until analysis. Processing involved the extraction of total RNA extraction and its retrotranscription to cDNA with PCR being carried out in a final volume of 20 μL. PCR and detection were performed using the MUC5AC detection kit following the manufacturer's instructions. The PCR products were visualized on PCR strips and in agarose gels.

The procedure was performed as previously described by Garcia et al. ²² Both sides of each disc were placed in contact with the epithelium using sterile tweezers to obtain the highest possible number of cells. One disc was placed on the corneal epithelium and a separate disc was placed on the bulbar conjunctival epithelium when sampled. Slight pressure was applied to the discs for a few seconds to improve sampling efficacy. The discs were immediately placed in an RNA protecting buffer (RNAprotect Cell Reagent; Qiagen, Valencia, CA) and stored at 4°C until use. 

RNA Isolation, Quality Assessment, and Reverse Transcription

Total RNA was isolated using a commercial RNA kit (RNeasy plus micro kit; Qiagen). Quantification and quality assessment of total RNA was performed using a microfluidics-based platform (2100 Bioanalyzer; Agilent Technologies, Inc., Santa Clara, CA) and RNA pico kits, using 1 μL of each sample (2 replicates for each sample). Reverse transcription of 10 μL of total RNA to cDNA was carried out using a cDNA kit (Transcriptor First Strand cDNA Synthesis Kit; Roche Diagnostics Deutschland GmbH, Mannheim, Germany) as previously described. ²² Negative PCR controls consisted on the inclusion of RNAse-free water in RT and PCR instead of a template. No limitation on sample RNA concentration or integrity (RIN) was established for inclusion in the study. 

Polymerase Chain Reaction and Detection

Ten μL cDNA were used as a template for all PCR reactions. PCR and detection reactions were performed by using the kit for the detection of MUC5AC expression in corneal cDNA (Limbokit; Operon S.A., Zaragoza, Spain), following the procedure described by the manufacturer. This kit includes all the reagents for PCR reactions (PCRmix, custom primers for amplification of MUC5AC and GADPH), and GoTaq Flexi DNA Polymerase, as well as reagents for the detection of amplification bands on the reverse-blot PCR strips. Briefly, 7 μL PCR-premix, 2 μL primers, 1 μL Taq, and 10 μL cDNA were mixed per reaction. The amplification conditions were: 96°C, 3 min/[96°C, 15 seconds; 60°C, 15 seconds; 72°C, 15 seconds] 40×/4°C on hold. The resulting amplicons were detected using the reverse-blot PCR-strips as described by the manufacturer (Operon S.A.) and an automated dispensing system (Autolipa; Innogenetics NV, Gent, Belgium). The system automatically performed all the steps for the detection with PCR-strips including: denaturation, hybridization, washing, and developing. All steps were performed at 42°C. The PCR products were also visualized by gel electrophoresis on 2.5% agarose-TAE gels. The results of amplicon detection by PCR strips versus 2.5% TAE-agarose gels were compared and the sensitivity of the reverse-blot PCR-strips was tested. 

Statistical Analysis

Statistical analysis was performed using statistical software (SPSS 19.0; IBM SPSS Statistics, Chicago, IL). The correlation between the results obtained with the PCR-strip for detection of the MUC5AC transcript in corneal epithelium and clinical diagnosis was evaluated using the Spearman correlation test. The level of statistical significance was considered to be P < 0.01. The sensitivity and specificity values of the technique were analyzed. The positive predictive value (PPV) was calculated by analyzing the number of true positives/(number of true positives + number of false positives) and the negative predictive value (NPV) was defined as the number of true negatives/(number of true negatives + number of false negatives). 

Results

To facilitate the interpretation of the results, an example of results using a TAE-agarose gel and a PCR-strip is illustrated (Fig. 2), with a positive and a negative sample for MUC5AC detection. 

Figure 2

 

Representation of MUC5AC amplification results obtained using gel (A) and PCR strip (B). Amplification of the MUC5AC transcript in the presence or absence of GAPDH is an indication of positive diagnosis for LSCD. The absence of MUC5AC with the presence of GAPDH is an indication of negative diagnosis for LSCD. The absence of both MUC5AC and GAPDH is considered a nonvalid result and is denoted as “inconclusive.” (A) 2.5% TAE agarose gel. Lane 1: Molecular weight markers. Lanes 2 and 3: positive results for MUC5AC; LSCD in cornea. Lane 4: amplification of GAPDH band only (i.e., a negative result [healthy cornea]). Lane 5: no amplification of either MUC5AC or GAPDH bands; negative control (dH₂O). (B) PCR-strip results. Lane 1: nondeveloped strip; only control line positions are observed. Lanes 2 and 3: positive results for MUC5AC. Lane 4: presence of GAPDH and control developing bands only; negative result for MUC5AC. Lane 5: appearance of control developing band only (i.e., inconclusive result). The developing control band is only visible after a correct developing of the strip. (C) Representation of the PCR-strip principle. The multiplex reaction mixture is subjected to PCR, and the amplification products including MUC5AC and GAPDH biotin-labeled amplicons are hybridized to their respectively probes that are anchored to the strip. After washing, a conjugate buffer is added and incubated. Finally the substrate is added and incubated until the appearance of bands. The reaction is stopped by washing with distilled water.

Molecular Diagnosis of LSCD by the PCR-Strip System Versus Clinical Diagnosis

An evaluation of the PCR-strip system for amplification of the MUC5AC transcript and detection of bands by reverse dot blot strips was performed in two studies analyzing 87 corneal epithelium samples collected by IC. Assessment of its value for diagnosis of LSCD was tested using 45 clinically diagnosed samples, and the specificity of the system was evaluated using 42 clinically healthy corneal samples. 

In the first study, a total of 34 corneas (19 LSCD and 15 healthy), and two conjunctivas from 22 patients were tested. The results of this assay are presented in Table 3. The PCR-strip system showed 20 positives (LSCD), 19 from LSCD patients and one from a control subject; 13 negatives (healthy corneas); and one inconclusive sample. Both conjunctivas analyzed and used as positive controls were positive for MUC5AC gene transcript. The correlation between the PCR-strip system and clinical diagnosis was 94.4% (P < 0.001). 

Table 3

 

MUC5AC PCR-Strip Test Detection Versus Clinical Diagnosis of LSCD

MUC5AC PCR Strip	Clinical Diagnosis of LSCD			CJ
	+	−	Total
Study 1
+	19	1	20	2
−	0	13	13	0
NC	0	1	1	0
Total	19	15	34	2
Correlation	19/19	13/15	32/34	2/2
Overall correlation			94.4%
Study 2
+	25	3	28	16
−	1	21	22	0
NC	0	3	3	0
Total	26	27	53	16
Correlation	25/26	21/27	46/53	16/16
Overall correlation			89.9%

 NC, not conclusive.

The results obtained by the PCR-strip system were confirmed using TAE agarose gels. This comparison indicated that all 36 samples tested showed exactly the same results in strip and gel (100% correlation). Although a weakly amplified band could be observed in the gel from one of the samples (sample A4), this band is longer in size than the MUC5AC band and was not observed in the strip, suggesting nonspecific amplification. A representative illustration of the results obtained is presented in Figure 3. 

Figure 3

 

Representative example of results obtained from samples tested for MUC5AC amplification in study 1. (A) Visualization of amplification bands on the PCR strips. (B) Visualization of same amplification products in a TAE 2.5% agarose gel. The improvement in the PCR-strip results is illustrated by sample A4. The arrow in (B) indicates the weak band that could be observed in the gel at a higher position than that corresponding to the MUC5AC band, whereas in (A), the arrow indicates the hypothetical position on the strip. The hybridization process and thorough washing of the strip rules out the presence of unspecific PCR product.

In this study, an additional comparison of results obtained with the PCR-strip system and the standard PCR technique as reported by Garcia et al. ²² was performed using the same corneal epithelium cDNA samples under exactly the same experimental conditions. The results indicated a correlation of 97.2% (P < 0.001) between the two techniques (35 from 36 samples with the same result), and revealed that the PCR-strip system was slightly more sensitive than the standard PCR technique (data not shown). 

In the second study, a total of 53 cornea samples (26 LSCD and 27 healthy) and 16 conjunctivas from 33 subjects were tested using the PCR-strip system procedure only. 

The results confirmed that 25 of 26 samples clinically diagnosed as LSCD were positive for MUC5AC, and one sample was negative for MUC5AC. Similarly, 21 of the 27 samples clinically classified as being healthy (no LSCD) were found to be negative for MUC5AC, three were found to be positive, and three were inconclusive because insufficient sample was collected by IC. All 16 conjunctivas were found to be positive for MUC5AC. These data indicate a correlation of 89.9% between the PCR-strip system and clinical diagnosis (Table 3). 

Finally, the results obtained by PCR strip were confirmed in TAE agarose gels (Fig. 4). This comparison indicated that, in general, no divergent results are observed in gel versus strips. However, clear differences in sensitivity and specificity were observed in 10 samples. In all those samples, better results were always obtained on the strips. Two samples showed a MUC5AC band, but did not present a GAPDH band in the agarose gel. However, a clear signal band for both MUC5AC and GAPDH was observed in the corresponding strips (samples indicated by dotted arrows in Fig. 4). In 3 cases, the GAPDH band was clear, but the MUC5AC band signal was very weak in the gel; however, clear bands were observed in strips (indicated by single asterisks in Fig. 4). 

Figure 4

 

Results obtained in study 2 for MUC5AC amplification. Bands obtained in the PCR strips (top) and in agarose gels (bottom) are shown. For ease of interpretation of results in the PCR strips, a pattern of bands is provided. In agarose gels, the MUC5AC and/or GAPDH bands are presented when appropriate. Dotted arrows indicate those samples in which only the MUC5AC band was visualized in agarose gel, whereas in the PCR strips, both MUC5AC and GAPDH could be seen in the same samples. Single asterisks indicate those samples in which the GAPDH band was clearly apparent, but MUC5AC was very weak in the gel; however, they were clearly positive in PCR strips. The bold arrow indicates a sample (154-C-OD) that was clearly negative in the PCR-strip, while weak and nonspecific amplification was observed in the agarose gel. This finding highlights the importance of the specific detection of the PCR strips, provided by the specific probe anchored to the strip membrane. Sample 094-CJ-OD (indicated by a triangle) illustrates a case in which MUC5AC amplification can be observed in the absence of the GADPH band; the result is considered MUC5AC positive. This sample also showed a very weak MUC5AC amplification band in the gel, while the PCR-strip indicated a clear MUC5AC band. Squares highlight the comparable areas in both gel and strips. C-OD, cornea-right eye; CJ-OD, conjunctiva-right eye; C-OS, cornea-left eye; CJ-OS, conjunctiva-left eye; CL, control line position; DC, developing control band; U, inconclusive result; N, PCR negative control consisting of the addition of H₂O instead of cDNA template in PCR.

Additionally, weak band signals for both GAPDH and MUC5AC were observed in gel for three samples, but clear bands were seen on the corresponding strips (double asterisks, Fig. 4). These samples represent the cases in which the increased sensitivity provided by the strip is crucial, because MUC5AC amplification is hardly detectable in gel. It was also observed that one sample (094 CJ-OD) did not present GAPDH in the gel nor in the strip; this sample showed very weak MUC5AC amplification band in the gel, while it showed clear MUC5AC amplification in the strip (indicated triangle in Fig. 4). Given that MUC5AC expression was detected in this sample, it was considered positive regardless of GAPDH amplification. Finally, one healthy sample (154 C-OD) seemed to show weak MUC5AC amplification in the gel, while no band was observed in the strip (indicated by a bold arrow in Fig. 4). This points to nonspecific amplification of an amplicon that is similar in length to the MUC5AC amplicon. This could lead to a false positive if only gel results were considered. All these observations indicated that PCR-strips enhanced the visualization of results, and thus represent a more sensitive technical format than agarose gels for viewing and evaluating results. 

Finally, a joint analysis of the results obtained in studies 1 and 2 indicated a global correlation between the PCR-strip system and clinical diagnosis of 91.4% (Table 4), since 44 of 45 samples clinically diagnosed for LSCD were positive for MUC5AC amplification, and only one was negative. In the healthy corneal samples included as negative controls, 34 of the 42 samples analyzed were negative for MUC5AC amplification, four healthy corneas were found to be positive for MUC5AC, and four samples were inconclusive because of insufficient sample in the corneal IC. All 18 conjunctivas used as positive controls were MUC5AC positive, ensuring the reliability of the results. Negative controls of PCR reactions, including the use of H₂O instead of cDNA, were processed under exactly the same experimental conditions. There was no amplification of either MUC5AC or GAPDH in these samples, thus confirming the absence of any contamination and the specificity of the reactions. The amplification of the GAPDH housekeeping gene in all healthy corneal epithelium samples confirmed the integrity of the samples, the appropriateness of PCR conditions, and the reliability of the negative MUC5AC results. 

Table 4

 

Global Correlation, Sensitivity, and Specificity of the MUC5AC PCR-Strip Detection System Versus Clinical Diagnosis of LSCD

MUC5AC PCR Strip	Clinical Diagnosis of LSCD			CJ
	+	−	Total
+	44	4	48	18
−	1	34	35	0
NC	0	4	4	0
Total	45	42	87	18
Correlation	44/45	34/42	78/87	18/18
Overall correlation			91.4%
Sen = TP/(TP + FN)			98%
Spe = TN/(TN + FP)			89%
PPV = TP/(TP + FP)			92%
NPV = TN/(TN + FN)			97%

 FN, false negatives; FP, false positives; Sen, sensitivity; Spe, specificity; TN, total negatives; TP, total positives.

According to the results obtained using the PCR-strip system compared with clinical diagnosis, the sensitivity of the PCR-strip for MUC5AC detection as an indication of LSCD in corneal epithelium was 98% and its specificity was 89%. The calculated positive predictive value (PPV) was 92% and negative predictive value (NPV) was 97% (Table 4). 

One of the samples in study 2 was used for a final study of the clinical potential of this PCR-strip system; we evaluated the possibility of an autologous limbal transplantation in an 18-year-old patient with LSCD in the left eye. LSCD was due to eye caustication, and this eye had previously been treated with amniotic membrane. The right eye was proposed as a limbal donor candidate for autologous transplantation, and MUC5AC expression was tested in both eyes prior to surgery (Fig. 5). Impression cytologies were taken from the cornea and conjunctiva of both eyes and processed for a reverse dot blotting PCR-strip. The test confirmed the presence of MUC5AC transcript in the damaged (left) eye cornea, and the absence of this transcript in the donor (right) eye. Conjunctival samples were used as positive controls for MUC5AC amplification. Based on these results, the candidate donor eye was confirmed to be normal, thus verifying the absence of involvement in the original injury, and further corroborating the viability of autologous limbal transplantation in this patient. 

Figure 5

 

MUC5AC detection in a patient with LSCD in the left eye due to corneal caustication. The right eye was proposed as a candidate donor for autologous limbal transplantation. MUC5AC transcript was tested in both eyes. (A) Donor candidate (right eye). (B) Affected (left eye). (C) Corneal and conjunctival IC samples were taken and processed under exactly the same conditions. (C) PCR-strip revealed the absence of MUC5AC in the candidate donor cornea (093 C-OD), confirming the healthy condition and viability of limbus transplant and confirmed the presence of MUC5AC in the affected cornea.

Discussion

LSCD has traditionally been diagnosed on the basis of patient history coupled with clinical observation of corneal conjunctivalization associated with persistent epithelial defects, which strongly hint at limbal stem cell deficiency. ²⁷ PAS-hematoxylin staining of corneal epithelium samples for the detection of goblet cells also facilitates the diagnosis. However, the fact that the latter provides limited sensitivity for diagnostic purposes, together with the unavailability of objective diagnostic tests, points to the necessity to develop more sensitive diagnostic methods. 

In an earlier study, we reported that the conventional PAS hematoxylin technique had a sensitivity of 37%. This contrasted dramatically with the 97% sensitivity estimated for the molecular diagnosis of LSCD involving direct MUC5AC transcript detection in IC samples. Other highly specific techniques for the diagnosis of LSCD, such as the immunohistochemical and biochemical characterization of proteins, including MUC5AC and several keratins and proteins (e.g., ΔNp63), have been reported. ^{4,11,13,14,16} However, these techniques are likely to be less sensitive than molecular detection based on PCR amplification, particularly in samples in which tissue is scarce, such as in IC samples. 

Highly sensitive techniques combined with IC are required to accurately diagnose LSCD, because erroneous diagnosis can lead to the patient undergoing unnecessary surgery. ²⁸ Confocal microscopy can be used for LSCD diagnosis, ^29–33 but the equipment required is rather expensive (not feasible for all centers) and a highly qualified professional is needed to correctly interpret the resulting images, thereby limiting its use. The present study evaluated the use of a novel technique for diagnosing LSCD that involves the detection of the MUC5AC transcript in corneal epithelium based on a reverse-blot PCR-strip. This system enables detection and visualization of the amplified MUC5AC fragment as a colorimetric band on a strip. We report here an overall correlation between the reverse-blot PCR-strip system and clinical diagnosis of 91.4%, demonstrating the high value of the system for MUC5AC transcript detection in corneal epithelium for diagnostic purposes. 

Discrepancies between molecular and clinical diagnoses were found in five samples. In study 1, the difference between clinical and molecular diagnosis was observed in one healthy cornea which expressed MUC5AC. The clinical history of this patient was consulted and confirmed the healthy condition. It is thus most likely that this false positive result may be due to imprecise sample collection. When the membrane for the IC was placed on the cornea, it could have touched the conjunctiva and been contaminated by goblet cells, thus producing a false positive. Since the reverse blot PCR-strip system involves transcript amplification based on polymerase chain reaction, it is a highly sensitive procedure in which a minimum quantity of goblet cells can lead to a positive result. This is a key advantage of this method, particularly when working with small samples of corneal epithelium. However, it is precisely this feature which makes correct sample collection a crucial step in the procedure. 

In study 2, differences between clinical and molecular diagnoses were found in four corneal samples. Three of them were clinically healthy corneas in which MUC5AC transcript was detected. All three samples were collected in the same hospital center, and a review of the corresponding clinical records confirmed their normal condition. Two of these samples were taken from the same volunteer diagnosed with dry eye, treated with artificial tears, and not submitted to any surgery. The third sample was taken from the left eye of a 58-year-old patient with LSCD in the right eye. This patient had a history of chronic inflammation in the left eye since childhood, but no clinical evidence of LSCD was apparent. Thus, this may be a case of subclinical LSCD with clinical signs that have not yet become apparent. Thus, the molecular method reported herein may be useful for the early detection of LSCD. 

The fourth incongruous sample was a corneal sample clinically diagnosed with LSCD; however, no MUC5AC amplification was evident. Detection of GAPDH in the sample confirmed the correct processing of the sample. This patient was resampled and retested, but MUC5AC expression continued to be absent. This patient presented peripheral corneal pannus only, which was thought to be due to LSCD with stable corneal epithelium in the central area. Thus, if the corneal IC has been collected from a more central area of the cornea, affected tissue would not have been obtained in the sample, thus explaining the absence of MUC5AC in this LSCD positive case. Alternatively, this case may involve a peripheral corneal pannus induced by causes other than LSCD. 

Four samples provided an inconclusive result (no MUC5AC or GAPDH amplification) likely due to insufficient sampling by IC. All inconclusive samples were from healthy corneas; one was found in study 1 and three in study 2. The RNA concentration and RIN value of the sample tested in study 1 were 215 pg/μL and 7.15 respectively, which are satisfactory values for amplification purposes. RNA degradation after quantification could therefore account for these results. The other three inconclusive PCR samples showed very low RNA concentrations (<60 pg/μL) and medium RINs (RIN <6). In total, there were 4/105 inconclusive RT-PCR results, and this represents only 3.8% of the samples. In such cases, we recommend resampling by corneal IC and retesting MUC5AC expression. Interestingly, only clinically healthy samples produced inconclusive RT-PCR results. This reflects the fact that more cells are usually gathered from conjunctivalized (LSCD) corneal epithelia than from healthy corneal epithelia, ¹³ as also demonstrated by RNA quantification using the Bioanalyzer (1.23 ng/μL vs. 0.65 ng/μL on average; P = 0.058). Although the concentration and integrity of all samples were analyzed in this work, the use of GAPDH as a control for sample quality and quantity on the strip makes it unnecessary to perform this step in the final protocol. The presence of the GAPDH band is crucial in negative samples, where the absence of MUC5AC in the presence of GAPDH must be obtained; otherwise, a false negative result can occur. However, in LSCD positive samples, MUC5AC (+) and GAPDH (−) results can also be obtained. This type of result can occur since the system is optimized to favor MUC5AC transcript amplification over GAPDH amplification, thereby enhancing MUC5AC amplification, even in samples with a very low concentration of RNA. 

Another important finding of this study is that the PCR strip detection system proved to be more sensitive than TAE-agarose gels for detecting amplification of MUC5AC amplicons from corneal epithelium samples, and thus it represents a more sensitive diagnostic tool. Whereas some samples showed weak signal for MUC5AC and/or GAPDH signals on gels (leading to possible misinterpretation of the results), these same samples showed clear bands on PCR strips, allowing a more objective and straightforward interpretation of results (as can be seen in Fig. 4). In contrast, weak nonspecific amplification bands representing false positives can in principle be observed at the MUC5AC position in the gel, but these are not observed on the strip after hybridization and washing. The hybridization of amplified products to specific probes in the membrane, the washing steps, and the specific streptavidin-biotin system involved in the developing step, ensure enhanced specificity in the PCR strip. 

The sensitivity (98%), specificity (89%), and positive (92%) and negative (97%) predictive values associated with the detection of the MUC5AC transcript by RT-PCR were satisfactory for LSCD diagnostic purposes, thus confirming the potential of this system for routine clinical diagnosis. Moreover, this diagnostic test is amenable to standardization involving the same experimental parameters in the diagnostic laboratories in different clinical centers. In comparison to the standard RT-PCR, the novel PCR-strip–based system is: (1) easier, given that a step-by-step, easy-to-follow protocol is presented, and the results are objective and straightforward; (ii) faster, because all reagents for PCR and visualization of results are precharged, thereby avoiding wasting time in preparing reagents, aliquots and premixes; (iii) cheaper, as the cost of equipment, reagents, and staff is lower; and (iv) easier to implement for routine use in clinic and diagnostic labs; the ophthalmologist has only to collect the corneal epithelium sample by IC, and send it to the diagnostic laboratory where the sample is processed. The final results, as strips with colorimetric bands, are communicated to the doctor for diagnosis/confirmation/monitoring of LSCD. 

Despite the high sensitivity of the test, there are some parameters related to sample collection and preservation that are fundamental for accurate results in diagnostic applications. These include correct corneal epithelium sample collection because very slight pressing of the membrane onto the cornea can lead to insufficient cell material for analysis. Alternatively, imprecise sample collection that involves accidentally touching the conjunctival epithelium with the membrane can render false positive results. Equally important is the preservation of RNA during sample collection and processing. Therefore, an RNA protection buffer is recommended for use immediately after sample collection, and working under RNase-free conditions during sample processing is indispensable to avoid RNA degradation. 

The system that we described here constitutes an easy method for the early detection of even mild cases of LSCD and for the corroboration of uncertain clinical cases. Moreover, this method has multiple potential applications beyond diagnosis. For example, it could be used to verify and monitor restoration of the corneal phenotype and the regression of goblet cells, ^34,35 monitoring of treated patients, evaluation of the evolution of the corneal epithelium after keratoplasty or after amniotic membrane transplant, and examination of the limbal condition of donor eyes before autologous stem cell transplantation (collection of limbal stem cells from a donor eye with subclinical LSCD may unbalance the candidate eye and induce or accelerate the onset of the disease). In addition, this method can be used for the monitoring of long-term soft lens users when regular follow-up is recommended. ²⁸ This could facilitate the early identification of focal LSCD, and subsequent cessation of wear may prevent the need for surgical intervention. 

This method also represents a useful tool for practitioners since it simplifies the selection of patients to be subjected to keratoplasty with a higher probability of success, thus reducing the risk of failure of this technique and also reducing unnecessary surgery. Based on the clinical experience that we have acquired during the present and previous studies, we have elaborated a decision tree which facilitates clinical praxis in cases of LSCD suspicion and possible therapeutic alternatives (Fig. 6). As presented in the suggested scheme, when any sign(s) or signal(s) of LSCD appears, a molecular diagnostic test is recommended to discard or confirm the pathological condition. When the molecular test yields a negative result, a regular penetrating keratoplasty (PK) or deep anterior lamellar keratoplasty (DALK) is recommended. In contrast, when a positive molecular result is accompanied by clinical confirmation of LSCD, evaluation of the contralateral eye of the patient should subsequently be performed. A negative result of the molecular test in the fellow eye indicates that it is a potentially suitable donor for autologous limbal transplantation or ex vivo expansion of limbal stem cells. In contrast, when a positive result is obtained confirming the implication of both eyes, other surgical alternatives such as allogenic limbal transplant, keratoprosthesis, or ex vivo expansion of autologous oral mucosal epithelial cells, ethmoidal mucosa, or mesenchymal stem cells may be indicated. 

Figure 6

 

Decision tree for LSCD management. The molecular diagnostic test reported here is recommended in cases in which any sign(s) or signal(s) of LSCD is apparent, when monitoring risk groups such as long-term soft lens users, or in situations in which surgical or treatment decisions require verification of the pathological condition. A negative result from the molecular test suggests that a regular PK or DALK procedure may be appropriate. In the presence of a positive result and confirmation of LSCD in the affected eye, the contralateral eye should be evaluated. A negative result of the molecular test in the fellow eye confirms that it can be a donor for autologous limbal transplant or ex vivo expansion of limbal stem cells. When a positive result is obtained in the fellow eye, other surgical alternatives, such as allogenic limbal transplant, keratoprosthesis, or ex vivo expansion of autologous oral mucosal epithelial cells, ethmoidal mucosa or mesenchymal stem cells should be indicated.

Finally, the use of this technique in clinical practice may have important economic repercussions since it can provide information about the indication and prognosis of corneal transplant in some patients diagnosed with LSCD. 

Acknowledgments

The authors thank Susana Gamen (Department of Research and Development at Operon) for her technical assistance. 

Supported in part by the Centre for the Development of Industrial Technology (CDTI) through its NEOTEC Program, Grant IDI-20080118. The authors alone are responsible for the content and writing of the paper. 

Disclosure: I. García, Bioftalmik (E), P; J. Etxebarria, None; J. Merayo-Lloves, None; J. Torras, None; A. Boto-de-los-Bueis, None; D. Díaz-Valle, None; R. Méndez-Fernández, None; A. Acera, Bioftalmik (E), P; T. Suárez-Cortés, Bioftalmik (E), P 

Footnotes

 IG and JE contributed equally to the work presented here and should therefore be regarded as equivalent authors.