Risk Factors for Recurrence of Postoperative Trichiasis: Implications for Trachoma Blindness Prevention

Background  Trachoma is the leading worldwide cause of preventable blindness. Surgerycan alleviate trachomatous trichiasis (TT), the blinding sequelae of trachoma,but recurrence rates are high.

Objective  To investigate risk factors for TT recurrence.

Methods  We conducted a prospective case-control cohort study in Nepal that includedpatients with TT who were undergoing bilamellar tarsal-rotation surgery andcontrol subjects who have scarring but no TT. Conjunctivae were graded fortrachoma and swabbed to detect Chlamydia trachomatis byin-house and commercial (Roche Amplicor; Roche Diagnostics Corp, Indianapolis,Ind)–polymerase chain reaction. Univariate and multivariate analyseswere performed for associations with chlamydiae and other risk factors.

Results  Postoperative TT recurrence rates were 11.1% (4 cases) at 6 months and25.0% (11 cases) at 12 months while no controls developed TT. Significantly,5 (45.5%) of 11 patients with TT recurrences at 12 months had chlamydial infectionat baseline (odds ratio, 6.0; 95% confidence interval, 1.5-24.3; P = .01), 8 (72.7%) at 6 months (odds ratio, 28; 95% confidence interval,4.3-181.0; P <.001), and 9 (81.8%) at 12 months(odds ratio, 48; 95% confidence interval, 7.5-302.6; P =.00). No controls were infected during this time. Active trachoma and thebaseline number of eyelashes touching the globe were also risk factors forTT recurrence.

Main Outcome Measures  The outcome variable was TT recurrence at 6 and 12 months.

Conclusion  To our knowledge, this is the first study to demonstrate that chlamydialinfection at the time of surgery and at follow-up is a significant risk factorfor postoperative TT recurrence.

Of the 38 million individuals who are blind in the world today, an estimated7 to 9 million (approximately 18%-24%) are blind because of trachoma,^1,2 a leading cause of preventable blindness.An additional 11 million have a complication of trachoma referred to as trachomatoustrichiasis (TT), that is, 1 or more eyelashes touching the globe or evidenceof epilation.^1,3 Trachomatoustrichiasis presents an immediate risk of blindness due to abrasion of thecornea, which frequently results in bacterial superinfection and healing byscar formation with the inevitable consequence of corneal opacification.^4,5 While the prevalence of trachoma hasdecreased in most of the developed world, the burden of blindness falls disproportionatelyin the developing world, where poverty, poor hygiene, and poor sanitationremain problematic.

Trachoma is a chronic ocular disease caused by the obligate intracellularbacterium Chlamydia trachomatis.⁶ Persistentor repeated childhood infection can lead to conjunctival scarring that occursas early as 5 years of age and can progress to more severe scarring and TTwithin 10 to 40 years.⁷ In hyperendemic areaswhere clinical disease and infection rates are high (30%-50%), 8% to 17.5%of the population may have TT.^8-11 TheWorld Health Organization (WHO) has initiated the Global Elimination of BlindingTrachoma by 2020, which calls for implementation of the SAFE strategy.¹² The SAFE strategy encompasses 4 different arms oftreatment and prevention of disease: (1) Surgery to prevent blindness in thosewho have trichiasis and/or entropion. (2) Antibiotics (tetracycline ointmentor azithromycin for active trachoma) to combat active chlamydial infection.(3) Facial cleanliness. (4) Environmental improvement.

Surgery is the most effective and direct means of treating TT. Manysurgical techniques have been used including cryoablation, electroepilation,eversion splinting, tarsal advance, tarsal grooving, tarsal advance and rotation,and bilamellar tarsal rotation.^13,14 Arandomized control trial in 1990 showed that the bilamellar tarsal-rotationprocedure was superior to other treatment methods because it had the lowestrecurrence rate regardless of coexisting eyelid closure defect, previous surgery,severity of trichiasis preoperatively, and the skill of the surgeon.¹³ It was also technically simpler and quicker thanthe tarsal advance and rotation method.¹³ Hence,the bilamellar tarsal-rotation procedure was recommended by WHO for TT surgery.¹⁵ However, TT recurrence has been reported with allsurgical techniques and has been observed from as early as 1 year to as lateas 11 years following the initial surgery.^8,14,16-18

The high rates of TT recurrence, up to 75%¹⁶ inmany trachoma-endemic countries worldwide, suggest that underlying environmentalrisk factors might contribute to disease progression. Patients with TT residein communities where reinfection is possible from household contacts or contactwith infected children. To our knowledge, the association between chlamydialinfection at the time of surgery and TT recurrence has not been studied. Asignificant correlation between infection and recurrence would support theneed for antibiotic therapy at the time of surgery to reduce and possiblyto prevent TT recurrence. Our study sought to evaluate whether infection with C trachomatis and other clinical factors at the time ofsurgery and at 6- and 12-month follow-ups would predict TT recurrence.

This was a prospective case-control cohort study designed to evaluaterisk factors for postoperative TT recurrence. The study was approved by theNepal Netra Jhoti Sangh (Nepali government prevention of blindness program),Kathmandu, and the institutional review board of Children's Hospital OaklandResearch Institute, Oakland, Calif.

A survey was conducted in the trachoma-endemic Kapilvastu District ofthe Lumbini Zone in Nepal to grade trachoma and identify cases of TT. Afterobtaining informed consent, research staff enumerated and clinically examinedall household members. Trachoma was graded using the modified grading scaleaccording to WHO guidelines¹⁹ by the same ophthalmologistat all time points as follows: follicular trachomatous inflammation (TF) (ie,the presence of ≥5 follicles ≥0.5 mm in diameter on the central uppertarsal conjunctiva); intense trachomatous inflammation (TI) (inflammatorythickening of upper tarsal conjunctiva with obstruction of >50% of the vessels);trachomatous conjunctival scarring (TS) (ie, the presence of easily visiblescarring in upper tarsus); TT (the presence of ≥1 eyelash touching theglobe or evidence of epilation); and corneal opacity (the presence of visiblecentral corneal opacity sufficiently dense to obscure the papillary margin).To classify the degree of TS, individuals were graded as having C1 for mildscarring, C2 for moderate scarring, or C3 for severe scarring.² Patientswere considered to have active trachoma if there was evidence of TF and/orintense TI.

Cases were defined as patients with individual eyelids having a trachomagrade of TT. Cases were matched by age and sex to controls, defined as thosewith conjunctival scarring without evidence of TT who resided in the samevillage. Visual acuity was assessed using a standard Snellen "E" chart measuredat 6 m and categorized using WHO definitions of visual impairment and blindness.Cases and controls were examined at baseline and again at 6- and 12-monthfollow-ups for TT recurrence in the operated on eye or for development ofTT in an unoperated on eye, respectively. Patients with TT were offered freecorrective eyelid surgery, which occurred at baseline.

To determine C trachomatis infection ratesfor both cases and controls, upper tarsal conjunctival swabs were collectedat baseline and at 6 and 12 months. Swabs were placed in collection media(M4-RT; Micro-Test, Lilburn, Ga) and labeled with a date and a unique identificationnumber. Specimens were maintained at −20°C before transport to thelaboratory where they were stored at −80°C until processed. Casesand controls with evidence of active trachoma were treated with topical tetracyclineaccording to WHO recommendations² after obtainingthe swab, although treatment was not observed. Sterile technique, includingthe use of a new set of surgical gloves for each case of TT, was performedto prevent cross-contamination of C trachomatis infectionamong patients. For controls, hands were washed with soap and water in betweeneach examination.

The bilamellar tarsal-rotation procedure was performed on all case ofminor (≤5 eyelashes touching the globe) and major (>5 eyelashes touchingthe globe) TT at baseline as recommended by WHO,¹⁵ includingpatients with TT with evidence of self-epilation. Each patient was reexaminedpostoperatively and at 6 and 12 months. Any surgical complications were noted.

All samples were processed according to the commercial (Roche Amplicor)–polymerasechain reaction (PCR) kit instructions (Roche Diagnostics Corp, Indianapolis,Ind) to detect C trachomatis in a masked fashion.The colorimetric assay contained positive and negative controls for each 96-wellplate; a 96-well internal control plate was processed in parallel with patientand control samples. The internal control plate served to identify inhibitorsfor PCR. When inhibitors were present, DNA was extracted from the originalsample and amplified using an in-house PCR protocol as described previously.²⁰ The final determination of a positive or negativeRoche Amplicor-PCR result was based on the results of the computerized platereader in comparison with positive and negative control readings as well aswith the respective internal control plate according to the package insert.To validate Roche Amplicor–PCR results, all equivocal and positive sampleswere subjected to in-house PCR testing. Samples were considered positive bythe in-house PCR when the product was the correct molecular weight in comparisonwith positive controls (and where negative control samples were negative)as determined by electrophoresis on a 2% agarose gel stained with ethidiumbromide.²⁰ Reinfection was defined as confirmedinfection at 2 different time points where an intervening sample was negative.Persistent infection was defined as positive infections at 2 or more consecutivetime points.

The outcome variable was the TT recurrence rate at 6 and 12 months asmeasured by 1 or more eyelashes touching the globe after surgery. The explanatoryvariable was C trachomatis infection status at eachtime point as measured by Roche Amplicor–PCR and in-house PCR results.Univariate and multivariate analyses were performed using Stata software (version7.0; Stata Corp, College Station, Tex). To maintain patient confidentiality,only the patient's individual identification number was used in the analyses.Contingency table analysis was performed to examine crude associations and χ² test or the Fisher exact test was used to determine significance.Multiple regressions were performed where the change in the β coefficientexceeded 10%. Patients with TT who were operated on and who had or did notexperience a recurrence for chlamydial infection were compared, and the associationof TT recurrence with the number of eyelashes touching the globe at the timeof surgery, the degree of TS, and the presence of active trachoma at the timeof surgery were analyzed. Cluster analysis was performed to compare exposureto infected household members for TT cases and controls as previously described.^21,22

Seventy-nine subjects were enrolled in the study; 53 (67.1%) had TTwith 1 eyelid operated on at baseline except for 1 patient who had bilateralsurgery. One patient with TT could not be operated on owing to alcohol intoxicationat the scheduled time of surgery. The sample size for analysis, therefore,consisted of 78 subjects with 79 eyelids (Table 1). All subjects from the community who had TT were enrolledin the study, and only 26 controls in the same community were available forage- and sex-matching with the cases. Of the patients with TT, 20 eyelids(25.3%) had minor TT and 23 eyelids (29.1%) had major TT at the time of surgery.Only 2 of the patients in the control group had active trachoma, both weregraded as having TF.

Of the samples that were positive by the Roche Amplicor–PCR testand by in-house PCR testing, all were of the correct molecular weight comparedwith positive control samples on agarose gels. One sample contained an inhibitorand tested negative by in-house PCR. The baseline infection rate of the patientswith TT was 24.5% (13/53); at 6 and 12 months the rates of TT were 27.8% (10/36)and 27.3% (12/44), respectively (Table 2); all controls tested negative for chlamydial infection at baselineand at 6 and 12 months. Cluster analysis of 171 (99.4%) of 172 household membersof the cases and controls demonstrated no difference for exposure to infectedhousehold individuals for those patients with TT who were operated on comparedwith controls. The average household size was 5 for cases and 5 for controlswith similar numbers of infected individuals among the households. There wasno household with only 1 individual. No statistical difference was noted inthe number of preschool children for households with cases compared with householdsfor controls. Infection for household members was assayed exactly as describedearlier; 26% were infected; there were no inhibitors detected for these samples.

There were no surgical complications. All cases of TT recurrence wereminor (≥5 eyelashes touching the globe); none of the controls developedTT during the 12-month period. At the 6-month follow-up, there were 4 casesof TT recurrence, constituting an incidence rate of 11.1%. Three of the 4cases had moderate scarring (TS C2) and 1 had mild scarring (TS C1) at baseline.At the 12-month follow-up, there were 11 cases of TT recurrence with a cumulativeincidence rate of 25.0%. Seven (63.6%) of these 11 were new cases, givingan incidence rate of 15.9%. While most controls had mild to moderate scarring(TS C1–TT C2) at baseline, 7 of the 11 recurrent cases had moderateto severe scarring (TS C2–TT C3) at 12 months.

Controlling for age and sex, univariate and multiple regression analysisshowed that infection with C trachomatis was associatedwith TT recurrence at 6 and 12 months (Table 3). Of the 4 cases of TT recurrence at the 6-month follow-up,2 patients (50%) were infected with C trachomatis atbaseline (P = .15), but all 4 patients (100%) wereinfected at 6 months (P = .001). Of the 11 casesof TT recurrence at the 12-month follow-up, 6 patients (54.5%) were infectedwith C trachomatis at baseline (P = .02), 8 patients (72.7%) were infected at 6 months (P <.001), and 9 patients (81.8%) were infected at 12 months (P <.001). An additional χ² test was performedto verify that those without follow-up resembled those at follow-up by sex,age, and infection status at baseline. We found that there were no significantstatistical differences between the 2 groups.

Persistent infection was also significantly associated with TT recurrenceat 12 months (Table 3). Six ofthe 9 cases of TT recurrence who were positive for infection at 12 monthswere also previously infected at 6 months, and 4 of these 6 patients had infectionat all 3 time points (P = .03 and P <.001, Fisher exact test, respectively). While we did not samplefor infection in between the follow-up time points of 6 and 12 months and,thus, cannot rule out reinfection, it is possible that these cases representpersistent infection. Odds ratio for association of TT recurrence with infectionwas statistically significant at 6 and 12 months (Table 4).

Multiple regression showed that the number of eyelashes touching theglobe at the time of surgery and the presence of an active trachoma infectionat 6 months were also risk factors of TT recurrence (Table 5). While active trachoma (TF/TI) at the time of surgery (baseline)was not predictive of TT recurrence, TF at 6 months was predictive of recurrenceat 12 months (P = .011) and intense trachomatousinflammation was predictive of recurrence at 6 and 12 months (P = .011 and P = .004, respectively). Visualimpairment, TS, and corneal opacity were not predictive of TT recurrence at6 or 12 months.

We know of no previous studies on the association of C trachomatis infection with TT recurrence following surgery. Our studyfound infection was a significant risk factor for postoperative TT recurrence.Recurrence has been well documented following surgery, but the rates tendto vary depending on the type of surgery, duration of follow-up, and countrywhere the surgery was performed. The bilamellar tarsal-rotation techniqueis considered to be the most effective operative procedure with the lowestrecurrence rates compared with tarsal advance, electrolysis, and cryoablation.¹⁴ In Oman, TT recurrence rates were reported to be20% to 23% for the tarsal rotation technique compared with 71% for electrolysisand 82% for cryoablation at 2 years.¹⁴ In Gambia,63 operated on eyes (55%) had recurrence at 10 years after tarsal rotationsurgery.²³ Similarly, Mali reported TT recurrencein 50 cases (75%) after 11 years.¹⁶ Recently,a study in Oman reported TT recurrence in 181 cases (61.8%) and 157 cases(50.6%) following tarsal rotation and electroepilation, respectively, overa follow-up period ranging from 1½ to 4 years.⁸ Ourstudy, reports a TT recurrence rate of 11.1% at 6 months and a cumulativerate of 25.0% at 12 months. Our finding of a significant association of C trachomatis infection with increasing rates of TT recurrenceover time provides one explanation for the observed recurrence rates notedin this and the aforementioned studies.

We found excellent correlation between infection and active trachoma,which is consistent with the theory that active trachoma may be a marker forchlamydial infection.¹¹ Munoz et al¹¹ found that while one third of the women with TF and37% to 44% of those with intense trachomatous inflammation had evidence forchlamydial infection, the rate of infection in women without active trachomawas low—4% to 5%. Another study found that 52.9% of those with TF and77.0% of those with intense trachomatous inflammation were positive for infectionby direct fluorescent antibody cytology.²⁴

Active trachoma at baseline and at 6 months correlated with the incidenceof TT at each time point in our study, but only active trachoma at 6 monthswas predictive of TT recurrence at 12 months. These findings are supportedby a study in Tanzania where, after controlling for age, the presence of activetrachoma among a cohort of women with TS was a significant predictor of the7-year incidence of TT.¹¹ In contrast, a studyin Gambia found that active trachoma was not associated with TT recurrenceat a median of 10 years after tarsal rotation surgery.²³ Itis possible that while active trachoma correlates with the prevalence andincidence of TT, once TT develops, 1 episode of active trachoma becomes lessimportant in recurrence. However, with ongoing or repeated episodes, whichcould indicate reinfection or persistence of infection, active trachoma mightbecome an important component of TT recurrence. Additionally, topical treatmentmay not be sufficient to interrupt this cycle. We found cases of reinfectionand persistence despite topical tetracycline treatment of active trachoma,which is consistent with previous studies where reemergence of infection wasnoted within 4 weeks of completion of topical treatment.^25,26 Thus,it will be valuable to conduct prospective studies that address the issueof whether specific prevalence rates of active trachoma correlate with bothinfection and higher rates of TT recurrence.

Herein, the significant association between infection and TT at baselinesuggests that infection is important for development of TT. We have also demonstratedthat reinfection or persistent infection is a risk factor for TT recurrenceafter surgery. Repeated infection has been shown to be important in the developmentof trachomatous scarring²⁷ and, therefore,the development of TT. Infection has also been shown to be significantly associatedwith the incidence of TT among women with scarring.¹¹ Theconcurrent presence of TS and exposure to chlamydiae may predispose patientsto reinfection and, therefore, progressive scarring.

Although none of the controls in this study had infection during the12 months of the study, they did have considerable TS, which reflects priorinfection. Only 2 of the controls had active trachoma, which is consistentwith a lack of active infection in this group as demonstrated by our PCR testresults. Based on cluster analysis, both TT cases and controls had equal exposureto infected household members, which strongly suggests that reinfection orpersistent infection in our population is linked with the pathogenesis ofdisease progression.

Loss to follow-up is a limitation of prospective cohort studies andwe encountered the same dilemma, especially at the 6-month follow-up. Presumably,patients who were doing well and free of postoperative TT recurrence weremore likely not to return for follow-up evaluation. However, 8 of those lostto follow-up at 6 months were seen again at 12 months, providing data at bothbaseline and 12 months. Our follow-up data at 12 months was 83% for TT casesand 100% for controls. Since we had sufficient numbers and statistical significanceat 12 months and since the association between infection and TT recurrencewas consistent at 6 and 12 months, this likely indicates that our findingsare valid for both time points.

The TT recurrence in our study was in the minor category, and likelydue to the short follow-up of 6 and 12 months. Other studies have found thatthe development of major TT after surgery increases with the time to follow-up.^8,14,18 For example, for caseswho underwent tarsal-rotation surgery in Oman, the rate of major recurrenceincreased from 15.4% at 1½ to 2½ years to 25.5% at 4 years.⁸ In Tanzania, the recurrence rate increased from 19%at 24 months to 21% at 36 months after tarsal-rotation surgery.¹⁸ Wewould, therefore, expect major TT recurrence to develop with time along withan increase in the overall rate of minor and major recurrence.

The number of eyelashes touching the globe at baseline was a risk factorfor recurrence at 12 months but not at 6. This may be due to the short durationof follow-up at 6 months. With shorter follow-ups, patients have less opportunityto become reinfected or to have progressive scarring that could result inrecurrence. Further, those with major TT at the time of surgery would be morelikely to have recurrence than those with minor TT as the underlying fibrosismay be more advanced. Since most of our TT cases at baseline had moderateto severe TS (TS C2-C3), it is not surprising that the number of eyelashesat baseline predicted recurrence at 12 months. Further, since there was noTT recurrence among those patients who had evidence of epilation at baseline,it is likely that there were even more cases of recurrence than what was observedbecause patients who had epilated in the past may also have been more likelyto epilate after surgery if they developed recurrence.

Although we had sufficient sample size for analyses in this study, largerprospective studies are required in other trachoma-endemic countries to substantiateour findings. Our results are important within the context of the SAFE strategyas they suggest the need for integration of surgery and antibiotic use topotentially prevent postoperative TT recurrence. Research on azithromycinor other treatment modalities at the time of surgery and at other postoperativetime points should be conducted to determine the most effective regimen forreducing TT recurrence.

Corresponding author and reprints: Deborah Dean, MD, MPH, Universityof California at San Francisco School of Medicine, Children's Hospital OaklandResearch Institute, 5700 Martin Luther King Jr Way, Oakland, CA 94609 (e-mail: ddean@chori.org).

Submitted for publication May 13, 2003; final revision received November19, 2003; accepted November 19, 2003.

This study was supported by a fellowship from Fight for Sight, New York,NY (Ms Zhang); grant ITI 01-040 from the International Trachoma Initiative,New York, NY (Dr Dean); and Public Health Service grant EY/AI12219 from theNational Eye Institute, National Institutes of Health, Bethesda, Md (Dr Dean).

We extend our deepest gratitude to the people of the Kapilvastu Districtof the Lumbini Zone, without whom this research would not be possible. Wealso thank the Director of Lumbini Rana-Ambika Eye Hospital, Amar Deuja, MD,for his tremendous assistance and support in conducting the fieldwork.