Utility of In Vivo Confocal Microscopy in Diagnosis of Acanthamoeba Keratitis: A Comparison of Patient Outcomes

Abstract

Purpose:

To compare outcomes between cases of Acanthamoeba Keratitis (AK) diagnosed and treated with or without the use of in vivo confocal microscopy (IVCM).

Methods:

We performed a retrospective comparative case series of 26 eyes of 23 patients diagnosed with AK at the Massachusetts Eye and Ear Infirmary over a 5-year period. The characteristics of all identified cases were summarized. We compared the time from presentation to diagnosis of AK (primary outcome), visual acuity, and rates of therapeutic penetrating keratoplasty (PK) between eyes diagnosed by culture only group (n=8) and by IVCM to diagnose AK (n = 9) and later confirmed by culture (IVCM/C group).

Results:

The diagnostic delay was significantly longer in the culture only group (25 ± 29 days) compared to the IVCM/C group (3 ± 3 days, p < 0.01). At 6 months, there was a significant difference in BCVA between the culture-only group (1.46 ± 1.07, n = 7) and the IVCM/C group (0.22 ± 0.22, n = 8), after adjusting for initial baseline VA (p = 0.02). Therapeutic PK was performed in 50% of culture only (n = 7), and 11% of IVCM/C group eyes (n=9), but this was not statistically significant (p = 0.13).

Conclusions:

IVCM can expedite the diagnosis of AK, and its use as an adjunct tool in the diagnosis of AK may result in better patient outcomes compared to basing treatment decisions on corneal cultures alone.

Background

Acanthamoeba is a ubiquitous protozoan found in water, air, and soil, which can cause a severe, painful, and sight-threatening keratitis. Risk factors include contact lenses wear, corneal trauma, or exposure to contaminated water or soil [1, 2]. The diagnosis of AK is difficult and often delayed due to the lack of specific clinical findings in the early stages. Empiric treatment of AK is usually not initiated without diagnostic confirmation, due to the high level of corneal toxicity of amoebicidal medications. The gold standard for AK diagnosis is to culture Acanthamoeba from corneal scrapings. However, Acanthamoeba can be difficult to culture, with positive culture rates ranging from 0% to 64% [2–8]. Moreover, even when cultures yield positive results, it may take up to 2 weeks for results to become available. In advanced cases, the organisms may be located deep in the corneal stroma, requiring corneal biopsy to obtain sufficient samples for culture. Complicating AK management further, the toxic effects of topical amoebicidal medications can be difficult to distinguish from active infection, potentially leading to unnecessarily prolonged treatment.

In vivo confocal microscopy (IVCM) has been used to image human corneas with high resolution, allowing the identification of larger microorganisms [9, 10]. Many authors have now reported on the ability to identify Acanthamoeba cysts and trophozoites by IVCM [11–16]. Several studies have demonstrated the high specificity and sensitivity of IVCM for the diagnosis of AK [14, 15, 17], which is comparable or superior to KOH prep or gram stain [18]. However, reports of patient outcomes using IVCM for the diagnosis and management of AK have to date been limited to case reports and non-comparative case series [2, 11, 13, 19]. The purpose of this study is to determine the potential advantage of using IVCM for AK by comparing outcomes in patients diagnosed with the aid of IVCM and patients diagnosed by corneal culture.

Methods

Institutional Review Board approval was obtained for this study, and this work was HIPAA-compliant. Patients diagnosed with AK at the Massachusetts Eye and Ear Infirmary (MEEI) Cornea Service between 2004 and 2009 were identified by reviewing microbiology laboratory and confocal microscopy databases of the Confoscan 4 (Nidek) and Heidelberg Retinal Tomograph 3/Rostock Cornea Module HRT3/RCM (Heidelberg Engineering) at the Ocular Surface Imaging Center. A retrospective review of patient charts and images was performed. The data were collected through July 2013. IVCM images had been initially reviewed at the time of patient evaluation for diagnostic purposes. For identification of cases for inclusion in this study, a single experienced observer (P.H.) later reviewed IVCM images of AK cases and other non-AK cases, while being masked to patient identifiers and clinical information. An example of a positive IVCM result, defined as the presence of one or more cysts or trophozoites, is shown in Figure 1. Only cases with positive confocal and/or positive culture or biopsy results were included. Patient records were excluded if fewer than 3 months of follow-up data were available.

Fig. 1. Imaging of Acanthamoeba cysts by Heidelberg Retinal Tomograph 3/ Rostock Cornea Module HRT3/RCM.

Round hyper-reflective objects with double-ring appearance are Acanthamoeba cysts in a patient with corneal stromal involvement (examples indicated with arrows).

For each patient, amoebicidal therapy had been initiated immediately after AK diagnosis was established by culture or IVCM. All eyes were treated with at least one cysticidal biguanide (polyhexamethylene biguanide [PHMB], chlorhexidine), but most eyes were treated with both of these agents. In addition, adjunct medications such as propamidine isethionate (Brolene), neomycin, gatifloxacin, and oral voriconazole, ketoconazole, or itraconazole, were often used by the treating physicians. Amoebicidal therapy was discontinued once there was clinical evidence of resolution for culture only eyes, or once there was no longer any evidence of AK on IVCM images for confocal group eyes in which serial IVCM was performed. Ameobicidal therapy was also discontinued after eyes underwent penetrating keratoplasty (PK), unless there was evidence of AK persistence or recurrence.

Patients were divided into two groups for analysis: cases diagnosed by culture results only and did not have IVCM performed (culture only), and cases with positive findings of Acanthamoeba on IVCM in addition to positive cultures (IVCM/C group). For comparisons of clinical outcomes, we excluded confocal cases that were not further diagnostically confirmed with a positive AK culture (from corneal or contact lens culture). Our primary outcome measure was the time between presentation at MEEI (including emergency ward visits) and AK diagnosis. Secondary outcomes were VA at initial presentation, 1 month, 3 months, 6 months, and final follow-up, and number and type of surgical treatments. Snellen BCVA measurements were converted into logarithm of the minimal angle of resolution (logMAR) VA prior to calculating the mean BCVA. Count fingers and hand motion VA were given logMAR VA assignments of 2 and 3, respectively, and light perception and no light perception values were excluded when calculating mean logMAR VA, according to the suggestion of Holladay et al. (2004) [20]. If BCVA had not been determined by spectacle refraction, the VA with pinhole, the best-available spectacle-corrected VA (cc), or uncorrected VA values (sc) were used in our calculations, in descending order of preference. The final BCVA was obtained from the most recent visit in which a manifest refraction had been performed after treatment of AK was complete, or if no manifest refraction was performed, then the available VA from the most recent visit was recorded. None of the corrected visual acuities used in our visual outcome analyses were with contact lens wear.

Publications were selected for comparison of outcomes by using a combination of the search terms “Acanthamoeba” and “keratitis” in PubMed (National Library of Medicine). All available PubMed abstracts and papers were reviewed. Studies comparable in design, presentation of outcomes and closest in time period to our study that did not use IVCM for diagnosis were included. Mean logMAR VA values were calculated for all studies where possible, and then converted into the nearest equivalent Snellen VA.

Statistical analyses were performed using Statistical Package for the Social Sciences v.28.0 (SPSS, Inc, Chicago, Illinois, USA). Mann Whitney U test was used for comparison of age and time to diagnosis between groups. Analysis of covariance (ANCOVA) was performed to compare VA changes at follow-up between groups after adjusting for baseline values. Wilcoxon rank-sum test was used to compare baseline and follow-up VA assessment in each group. Fisher’s exact test was used for the comparison of PK prevalence.

Results

Thirty-two patients diagnosed with AK were identified by reviewing microbiology and IVCM image interpretation records between 2004 and 2009. IVCM diagnosis was confirmed for all cases in a masked fashion by a single experienced observer (P.H.). Nine patient records were excluded due to missing data or follow-up less than 3 months. Of the remaining 23 patients (26 eyes), 8 eyes of 8 patients were diagnosed and treated based on culture results only, and 18 eyes of 15 patients were diagnosed based on positive Acanthamoeba identification by IVCM (total IVCM group). Of the 18 eyes having had IVCM-based diagnosis of AK, 5 eyes had positive Acanthamoeba cultures from corneal scraping at MEEI, 2 had positive contact lens cultures at MEEI, and 2 had a confirmed report of positive corneal culture from outside institutions. Thus, nine of the eyes diagnosed by IVCM were culture-confirmed (IVCM/C group). Patient demographics and risk factors are shown in Table 1. The IVCM/C and culture only (C) groups compared in our analyses of clinical outcomes were comparable in terms of age (p = 0.38) and follow-up duration (p = 0.44).

Table 1.

Patient Demographics

	Culture only group	Total IVCM group	IVCM/C group*
Total patients	8	15	8
Total eyes	8	18	9
Mean Age (years ± SD)	43 ± 21	35 ± 14	33 ± 17
Range (age in years)	13 – 69	13 – 59	13 – 59
Gender
Female (%)	3 (37.5%)	5 (33%)	3 (37.5%)
Male (%)	5 (62.5%)	10 (67%)	5 (62.5%)
Eye
Right	8	11	5
Left	0	7	4
Risk Factors
Contact lens (%)	6 (75%)	13 (87%)	7 (87.5%)
Corneal trauma (%)	1 (12.5%)	1 (6.7%)	0 (0%)

^*Positive Acanthamoeba from cornea or contact lens/case culture. IVCM = in vivo confocal microscopy, IVCM/C = Culture-confirmed IVCM. SD = standard deviation.

Tables 2 and 3 summarize the individual and group clinical outcomes of our culture only and IVCM/C cases. One of the 8 IVCM/C patients had bilateral involvement, and none of the 8 culture only group patients had bilateral disease.

Table 2.

Individual Outcomes

Category	Eye	Year of presentation	Corneal culture	Diagnostic Delay (Days)	Initial VA	1 month BCVA*	3 month BCVA*	6 month BCVA*	Final BCVA*	Surgeries	Follow-up (mo.)
Culture only	1	2005	Positive	10	20/400 sc	20/100 ph	CF 4 ft sc	CF 2 ft sc	CF @ 1 ft sc	Therapeutic PK × 3, graft failure × 3	97
	2	2004	Positive	95	HM 3 ft sc	20/70 ph	HM sc	HM sc	20/70 ph	Therapeutic PK	28
	3	2006	Positive	8	20/40 cc 20/30 ph	20/20 cc	20/20 cc	20/20 cc	20/20	None	39
	4	2006	Positive	12	CF 3 ft cc	20/60 ph	LP sc	CF 1–2 ft sc	20/600 (BCL)	Therapeutic PK × 2, graft failure × 1	85
	5	2006	Positive	29	20/40 cc	20/200 sc	20/60 ph	20/80	20/50	Therapeutic PK, CE/IOL	79
	6	2007	Positive	19	HM sc	CF 2 ft sc	CF 1 ft sc	CF 2 ft sc	CF sc	None	7
	7	2006	Positive	8	20/20 cc	20/20 cc	20/20 cc	Unavailable	20/20	None	44
	8	2006	Positive	20	20/500 cc 20/200 ph	CF 2 ft sc	20/70 ph	20/80 ph	20/30	None	16
Total IVCM group
	1	2008	Positive	5	20/400 sc	20/200 sc	20/30 ph	20/25	20/20	Therapeutic PK, CE/IOL	53
	2	2007	Positive	1†	20/80 cc 20/40 ph	20/40 ph	20/25 cc	20/20	20/20	None	35†
	3	2007	Positive	1†	20/50 cc 20/40 ph	20/30 ph	20/40 cc	20/40	20/30	PTK for scar	35†
	4	2008	PositiveΔ	6	20/150 cc	20/25 ph	20/20 ph	20/20 ph	20/20	None	23
	5	2006	PositiveΔ	4	20/30 cc 20/25 ph	20/25 ph	20/25 ph	unavailable	20/25	None	3
	6	2008	Positive	7	CF 2 ft sc	HM	CF 2ft cc	20/70	20/40	CE/IOL	37
	7	2007	Positive	0	20/100 sc	20/25 ph	20/100 cc	20/60 ph	20/15	PTK/PRK for scar	51
	8	2009	Positive‡	0	20/200 cc 20/30 ph	20/80 cc	20/30	20/40 cc	20/30	None	25
	9	2009	Positive‡	0	20/50 cc 20/30 ph	20/25 ph	20/20 cc	20/20 cc	20/40	Surgical biopsy	40
	10	2008	Negative	16	20/40 sc 20/20 ph	20/70 ph	20/50 ph	unavailable	20/25	None	31
	11	2006	Negative	2	20/50 cc 20/30 ph	20/100 ph	20/40	20/25	20/20	Therapeutic PK	78
	12	2008	Negative	1†	20/200 sc	20/20 sc	20/20 sc	20/20 sc	20/20	None	54†
	13	2008	Negative	1†	CF 4 ft sc	HM	20/40 sc	20/20 sc	20/40	Therapeutic PK, CE/IOL	54†
	14	2008	Negative	7†	20/50 cc	20/30 cc	20/20 cc	20/20 cc	20/20	None	25†
	15	2008	Negative	7†	20/60 cc 20/50 ph	20/30 cc	20/20 cc	20/20 cc	20/20	None	25†
	16	2008	Negative	0	CF 1 ft cc	20/80 ph	20/200 cc	20/200 cc	20/30	Combined therapeutic PK/CE/IOL, graft rejection × 1, repeat PK	55
	17	2007	Negative	4	20/25 cc	20/20 cc	20/25 cc	20/20 cc	20/20	None	46
	18	2008	Negative	2	20/50 cc 20/30 ph	20/20 cc	20/20 cc	unavailable	20/20	None	5

BCL = bandage contact lens, BCVA = best-corrected visual acuity, cc = with spectacle correction, CE/IOL = cataract extraction and intraocular lens implantation, CF = count fingers, HM = hand motions, ph = pinhole, PK = penetrating keratoplasty, PTK = phototherapeutic keratectomy, PRK = photorefractive keratectomy, sc = without correction, VA = visual acuity;

^*BCVA by manifest refraction unless otherwise annotated,

^†Bilateral cases,

^Δculture positive from contact lens/case only,

^‡culture positive by report from outside institution.

Table 3.

Comparison of outcomes between groups and published studies

	Culture only	IVCM/C	Ku et al.	Por et al.	Agla et al.	Butler et al.	Chew et al.
Data collection period	2004 – 2013	2006 – 2013	2003 – 2007	2000 – 2007	2000 – 2002	1997 – 2002	2004 – 2008
Country	USA	USA	Australia	Singapore	France	Australia	USA
Number of patients (eyes)	8 (8)	8 (9)	13 (13)	42 (43)	8 (8)	20	59 (59)
Mean months follow-up ± SD (range)	49 ± 34 (7–97)	33 ± 16 (3–53)	14 (3–53)	13 (0.2–94)	NS (4–14)	24.8 (1–72)	11 (0.4–35)
Culture/Histology positive (% patients)	8 (100%)	9a (100%)	12 (92%)	37 (86%)	8 (100%)	11 (55%)	36 (61%)
Mean days to diagnosis ± SD (range)	25 ± 29 (8–95)	3 ± 3 (0–7) (p<0.01)f	20 (7–84)	28.6 (4–174)	64 (8–240)	26.6 (0–115)	39 ± 49 (4–355)
Total PK or DALK (% eyes)	4 (50%)	1 (11%)	1 (8%)	15 (35%)	1c (13%)	6 (30%)	13 (22%)
Therapeutic (% eyes)	4 (50%)	1 (11%)	NS	12 (28%)	1 (13%)	5 (25%)	5 (8%)
Mean Visual Acuity
Initial VA	20/445	20/91 (p=0.16)f	20/100	NA	20/400	20/200	20/100
Final BCVA	20/141	20/25 (p=0.04)f	20/80d	NA	20/100	20/40	NA
Final Visual Acuity Ranges
20/40 or better	3 (37.5%)	7 (100%)	7 (54%)	25 (60%)	3 (37.5%)	15 (75%)	(~57%)e
20/50 to 20/200	2 (25%)	0 (0%)	2 (15%)	11 (26%)	3 (37.5%)	4 (20%)	(~15%)e
Worse than 20/200	3 (37.5%)	0 (0%)	4 (31%)	6 (14%)	2 (25%)	1 (5%)	(~28%)e

NA, not available; PK, penetrating keratoplasty; DALK, deep lamellar keratoplasty; IVCM/C = Culture-confirmed IVCM. SD, standard deviation;

^aincluding contact lens and case cultures,

^breported in this study as time to sampling for culture,

^cnot including 2 planned PKs in Agla et al.,

^dnot including one NLP eye,

^eapproximations based on visual assessment of graphic representation of data,

^fMann-Whitney U-test result compared to conventional.

For comparative analysis, we only compared clinical outcomes between culture only and the subset of total IVCM cases that were also cornea or contact lens culture-confirmed eyes (IVCM/C). The mean days to diagnosis from initial presentation to MEEI was significantly longer (25 ± 29 days) in the culture only group compared to 3 ± 3 days in the IVCM/C group (p < 0.01). The mean initial logMAR VA was 1.35 ± 1.21 (n = 8) in the culture only group, and 0.66 ± 0.64 (n = 9) in the IVCM/C group, and was not significantly different (p = 0.32). When adjusted for baseline VA, mean BCVA was not significantly different in the culture only group (0.84 ± 0.79 (n = 8)) compared to the IVCM/C group (0.61 ± 1.07 (n = 9)) at 1 month (p = 0.77), or at 3 months, (1.15 ± 1.18 (n = 7) vs. 0.40 ± 0.63 (n = 9), respectively, (p = 0.30)). However, at 6 months, there was a significant difference in BCVA between the culture only group (1.46 ± 1.07, n = 7) and the IVCM/C group (0.22 ± 0.22, n = 8), again when adjusted for baseline VA (p = 0.02).

In the culture only group, no significant difference was observed between baseline and 1 month (p = 0.27), 3 months (p = 0.58), 6 months (p = 0.69), and final VA assessment (p=0.13). In contrast, in the IVCM/C group there was no significant difference between baseline and 1 month (p = 0.32) and 3 months (p = 0.43) BCVA, but there was significant improvement at 6 months (p = 0.03) and at the final follow-up (p = 0.03) when compared to baseline.

Two eyes in the culture only group had a final BCVA worse than their initial VA, whereas none of the IVCM/C eyes had a final BCVA worse than their initial VA. At 6 months, 20/40 or better BCVA was achieved by 14% of culture only (n = 7), and 75% IVCM/C (n = 8) eyes. The percentage of eyes with 20/40 or better final BCVA was 37.5% in the culture only group and 100% in the IVCM/C group. In comparison, between 37.5% and 75% of culture only eyes in previously published studies had VA better than or equal to 20/40 at final follow-up (Table 3) [4, 21–25].

When only eyes with initial VA of 20/100 or worse were compared, in order to select for cases with potentially more advanced disease at presentation, the mean logMAR BCVA was significantly better in the IVCM/C group (n = 4) than in the culture only group (n = 5) at 6 months (0.28 ± 0.29 versus 1.92 ± 0.86, respectively, p = 0.013), and at the final follow-up (0.1 ± 0.1 versus 1.3 ± 0.8, respectively, p = 0.026), after adjusting for baseline VA values. When comparing eyes that had initial VA of 20/50 or better, in order to select for cases with potentially less severe or early disease at presentation, the mean BCVA was not significantly different between the IVCM/C group (n = 3) and the culture only group (n = 5) at 6 months (0.15 ± 0.2 vs 0.30 ± 0.42 logMAR, respectively), and at the final follow-up (0.15 ± 0.1 vs. 0.16 ± 0.2 logMAR, respectively), again after adjusting for baseline VA.

Therapeutic penetrating keratoplasty (PK) was performed in 4 of 8 culture only eyes (50%), and 1 of 9 IVCM/C eyes (11%). This difference was not statistically significant (p = 0.13). Individual surgical interventions are shown in Table 2.

Discussion

To our knowledge, this is the first study to directly compare outcomes among AK patients diagnosed and treated with or without the use of IVCM as a diagnostic tool. We found that there was a significantly shorter time from presentation at our institution to diagnosis of AK in patients imaged by IVCM. In addition, we also found that patients imaged by IVCM to aid in the diagnosis of AK, but also confirmed with a positive AK culture, had generally better visual outcomes compared to patients diagnosed by culture only, even after adjusting for baseline VA. Further, the IVCM/C group showed a significant improvement in VA between baseline and follow-up, whereas no significant difference was observed in the culture only group, further suggesting an advantage to using IVCM to detect AK. We additionally found that in cases where the vision was severely impaired at initial presentation (20/100 or worse), the IVCM/C group eyes had better visual outcomes. Whereas, when the initial vision was good at presentation (20/50 or better), both groups tended to have good final visual outcomes.

Prior reports of AK outcomes in small series of patients diagnosed with IVCM have shown good visual outcomes that were comparable to our results [11, 13, 19]. In particular, Matsumoto et al.[13] found that 3 patients improved to VA of 14/20 or better, with initial VA ranging from 12/20 to count fingers. Alomar et al.[11] reported outcomes of 2 patients with early stage AK with IVCM, who had VA of 20/20 and 20/30 initially, and final VA of 20/20 for both. In an earlier paper by Cho and Holland [19], outcomes in 4 patients diagnosed by IVCM demonstrated a mean final VA of 20/30 from a mean initial VA of 20/140.

The generally good VA outcomes in AK cases diagnosed by IVCM may be attributable to a more rapid diagnosis by IVCM compared to corneal cultures alone. The reported mean delay in diagnosis of 26.6 days by Butler et al.[4] was similar to the mean of 25 days in our culture only group, and substantially longer than that of our IVCM/C group (3 ± 3 days). In the other studies summarized in Table 3 [21–25], the time to diagnosis was measured as the onset of symptoms to diagnosis or as the onset of symptoms to the date of sampling for culture, which does not allow meaningful comparison to our data. In our study, we measured the time to diagnosis from presentation at MEEI, as our aim was to determine the speed of diagnosis by IVCM compared to culture and clinical diagnosis at our institution.

There was no statistically significant difference in the prevalence of therapeutic PK in the culture only compared to the IVCM/C group. Also, compared to published results for non-IVCM eyes (8–28%), the prevalence of therapeutic PK in our IVCM/C group (11%) was similar (Table 3). However, the lack of statistical significance may be due to the small group of cases reviewed in this study. Late diagnosis of AK has been associated with increased rates of PK [26]. Thus, it would be expected that if IVCM speeds AK diagnosis that there may be lower rates of PK in confocal-diagnosed eyes. Further studies will be needed to determine if any association with IVCM and reduced rates of PK exists.

IVCM operator and reader experience is known to be a major factor in determining the sensitivity and specificity of IVCM in diagnosing AK [27]. In our study, a single experienced IVCM reader (P.H.) reviewed all images, to confirm the diagnosis listed in our confocal database. Analysis of the sensitivity and specificity of AK diagnosis was not within the scope of this study, and has been discussed by others [14, 15, 17]. As Acanthamoeba can appear in cystic or trophozoite form in the cornea, and can vary in size, there is certainly potential for confusing Acanthamoeba organisms with other entities, such as superficial epithelial cell nuclei, inflammatory cells, or microcysts in Meesmann corneal dystrophy [28]. We therefore agree with Hau et al.[27] that IVCM is more powerful when read by an experienced observer.

Our study has several limitations, mostly attributable to the fact that it is a retrospective analysis of a relatively rare disease. The number of cases was relatively small. Given that this is a retrospective study of patients cared for by different physicians, there could not be a single standardized treatment protocol. With the availability of IVCM, patients suspected of having AK were potentially being referred sooner and with less severe disease. However, there was no statistically significant difference in baseline VA at presentation between the IVCM/C and culture only groups. Furthermore, differences in our outcomes were significant even after adjusting for baseline VA. Not all of the confocal group cases had positive cultures to confirm the diagnosis beyond clinical and IVCM findings, and we thus limited our secondary analyses on the culture-confirmed subset of confocal-diagnosed cases in order to reduce the potential for false positives, which reduced our sample size further. Finally, as the patient entry points into the study are separated in time based on when IVCM became available, it is possible that increased clinical experience or awareness of AK may have influenced the outcomes to some degree.

In summary, our study suggests that IVCM can help provide immediate and rapid diagnosis of AK and allow prompt institution of appropriate treatment, which may contribute to improved patient outcomes. Prospective studies and larger studies will be important to further determine the impact of IVCM use on AK patient outcomes.