Clinical differentiation of keratitis due to fungus and Pythium: A photographic survey

Abstract

Purpose:

To evaluate the knowledge and skills of clinicians in differentiating cases of Pythium from fungal keratitis based on their clinical appearance.

Methods:

Thirty slit-lamp photographs from culture-proven cases of Pythium and fungal keratitis were selected for conducting a prospective online-based photographic survey. The participants were asked to identify the causative organism and enlist the specific clinical features which helped in reaching the diagnosis. Both cornea specialists and clinical fellows/trainees participated in the survey. Sensitivity, specificity, and predictive values were estimated, and the factors associated with these, including the identification rates of clinical signs, were evaluated using bivariate and multivariate linear regression analyses. The identification rates between the consultants and fellows were compared.

Results:

A total of 42 cornea specialists (28 consultants and 14 clinical fellows) participated in the survey. Clinicians were able to differentiate Pythium from fungal etiology in only 56% of cases. The differentiating ability between the consultants and fellows was statistically insignificant. Average sensitivity, specificity, and positive and negative predictive values for differentiating Pythium from fungus were 56%, 65%, 56%, and 66%, respectively. The specific clinical features enlisted by the participants to differentiate Pythium from fungus were the presence of tentacles, peripheral guttering, dot-like infiltrates, and elevated plaques.

Conclusion:

Clinical knowledge regarding the signs and symptoms and the microbiological identification of features for Pythium are still lacking among clinicians. As the treatment protocols are distinctly different for both, it is imperative to enhance the knowledge and diagnosing ability to tackle this emerging microorganism which causes high morbidity.

Pythium insidiosum is considered to be a parafungus that causes keratitis similar to fungal keratitis. It is an oomycete with zygomycetes fungus-like features of branching and broad, sparsely septate, or aseptate filaments often with ribbon-like folds. It is recalcitrant to antifungal therapy because it lacks ergosterol in cytoplasmic membrane.[1] It can cause infections in both humans and animals, mostly in tropical and temperate countries. Even though corneal infections are rare, they could lead to vision-threatening complications.[2,3] Most of the reports of Pythium keratitis was primarily from Thailand,[4] but recently, similar cases from Australia, the USA, India, and Israel have been reported, thus indicating that the organism is more widely distributed and is probably underdiagnosed.[5-7] The prevalence of P. insidiosum as an etiological agent of keratitis in India varied between 5.5% and 3.9% according to a study by Sharma et al.[8] The study by Hasika et al.[9] from South India has also found the prevalence to be 5.9% over a period of 2 years, where they had diagnosed Pythium in 71/1204 of all keratitis cases that were culture positive for the fungus.

In temperate countries where fungal corneal infections predominate, it is very important to be vigilant about an organism that mimics fungal keratitis clinically, microbiologically, and histologically, but with a different line of treatment. Accurate diagnosis of Pythium keratitis requires DNA sequencing, which may not be available in all laboratories.[6-8] Treatment protocol for Pythium keratitis is novel and still evolving. Pythium keratitis progresses very rapidly and responds poorly to medical treatment.

If not treated properly, the majority of cases eventually require keratoplasty or evisceration.[9,10] Surgical outcomes of keratoplasty are also poor, with most grafts getting either reinfected or failed. Hence, early diagnosis and treatment are very important. Given the critical role in the therapeutic decision-making process, it is still uncertain how accurate are clinicians in predicting the causative organism based solely on appearance. Also, the precise clinical signs that can help in making a definite diagnosis of Pythium keratitis are unclear.

The purpose of the study is to evaluate the knowledge and skills of clinicians in distinguishing cases of Pythium from fungal keratitis based only on the clinical appearance.

Methods

After obtaining approval from the Institutional Ethics Committee (2021-73-BHR-37), a photographic survey was initiated via Google Forms platform, and the weblink was sent to a group consisting of corneal specialists (both consultants and clinical fellows/trainees). Thirty slit-lamp photographs from culture-proven cases of Pythium and fungal keratitis were selected from the electronic medical records database of the institute. All slit-lamp photographs used in the survey were taken by an experienced ophthalmic photographer using a CANON EOS30D camera that produces images (3504 pixels width and 2336 pixels height each) with a uniform resolution of 72 dpi (horizontally and vertically). The survey required an opinion to differentiate cases of Pythium from fungal keratitis. Graders were informed that 13 out of the 30 photographs were taken from eyes with culture-proven Pythium keratitis, and the remaining were from eyes with fungal keratitis. Duration of presentation (within 1 month) and size of keratitis were matched between the two etiologies during selection of the clinical pictures. The survey was created in a simple manner, so that all questions could be completed in <10 min.

After giving a formal consent, for each photograph, the participants had to indicate whether the corneal ulcer is most likely Pythium or not and should list the specific clinical sign most helpful in reaching their conclusion among the four given options: peripheral guttering, dot-like infiltrate, elevated plaque, and tentacles. We sent the weblink to all the participants on Feb 15, 2021, a second reminder was sent on Feb 24, 2021, and after a period of 3 weeks, we closed the survey. We defined a clinical sign as present for each photograph if it was listed by more than 50% of the participants as a diagnosing feature. The completed surveys were prospectively collated. We computed the fraction of photographs correctly identified as Pythium infection or fungal infection. Patient identifying information was not used.

The statistical analysis was performed using STATA v14.2 (StataCorp, College Station, TX, USA). Categorical data were described in proportions. Sensitivity, specificity, and predictive values were estimated in all the participants to distinguish Pythium from fungal etiology. Bivariate and multivariate linear regression analyses were used to evaluate factors associated with sensitivity, specificity, predictive values, and identification rates of clinical signs. The identification rates of clinical signs between consultants and fellows were compared by Mann–Whitney test. A P value of <0.05 was considered statistically significant.

Results

A total of 42 cornea specialists participated in the survey, which included 28 consultants and 14 clinical fellows. Among 28 consultants, seven (25%) had <5 years, 10 (35.7%) had 5–10 years, and 11 (39.3%) had >10 years of experience. All participants had access to the microbiological laboratory with a minimum of 5–10 cases of microbial keratitis being diagnosed per week in their respective institute(s). Although all the clinicians who participated in this survey were aware of Pythium keratitis and had good knowledge about the clinical features, 20 (47.6%) reported having diagnosed very few cases (less than five cases per year) in their career. Four of them mentioned having not seen a single diagnosed and microbiologically proven case of Pythium in > 5 years of their career as a cornea specialist.

Of the 30 clinical photographs, 13 were from culture-positive P. insidiosum cases and the rest were having fungus etiology [Fig. 1]. The causative organisms among fungal cultures were Fusarium spp. (n = 7, 41.2%) and Aspergillus spp. (n = 10, 58.8%). We evaluated the sensitivity and specificity of all the participants to distinguish Pythium from fungal etiology. These, along with positive predictive value (PPV) and negative predictive value (NPV), are summarized in Table 1. Whether the participant was a consultant or a clinical fellow, years of clinical experience and number of Pythium cases seen in their career did not have any effect on sensitivity, specificity, PPV, and NPV.

Figure 1.

Slit-lamp photographs of Pythium keratitis (a and b) and fungal keratitis (c and d) in diffuse illumination

Table 1.

Summary of sensitivity, specificity, positive predictive value, and negative predictive value in distinguishing Pythium from fungus

Summary	Sensitivity	Specificity	Positive predictive value	Negative predictive value
Mean±SD	55.9%±17.3%	64.4%±16.7%	55.7%±13.7%	66.2%±10.2%
Median	53.9%	64.7%	54.2%	66.7%
25th percentile	46.2%	54.4%	44.9%	58.8%
75th percentile	69.2%	70.6%	61.2%	71.4%
Minimum	23.1%	0.00%	30.0%	47.1%
Maximum	100%	94.1%	83.3%	100%
P
Consultant	0.73	0.96	0.67	0.97
Experience	0.60	0.46	0.54	0.83
Number of Pythium cases seen	0.49	0.44	0.25	0.42

SD=standard deviation

The specific clinical features enlisted by the participants to differentiate Pythium from fungus were the presence of tentacles, peripheral guttering, dot-like infiltrates, and elevated plaques. Table 2 shows the identification rates of the participants by these specific clinical features. The clinical signs most commonly used by the participants in identifying Pythium in photographs were tentacles (~26%) and peripheral guttering (~26%). Bivariate analyses found trainee status to be significantly associated with the identification of dot-like infiltrates, peripheral guttering, and tentacles. The analyses also showed years of experience to be significantly associated with the identification of dot-like infiltrates. Multivariate analysis for the identification of dot-like infiltrates found only trainee status to be significant and not the years of experience. This showed that years of experience and number of Pythium cases seen in career did not affect the identification of clinical features.

Table 2.

Summary of the Pythium identification rates by specific clinical features

Summary	Dot-like infiltrates	Peripheral guttering	Tentacles	Elevated plaque
Mean±SD	19.1%±14.2%	25.8%±13.4%	26.0%±14.8%	16.3%±14.7%
Median	19.2%	23.1%	23.1%	15.4%
25th percentile	7.7%	15.4%	15.4%	7.7%
75th percentile	30.8%	36.5%	36.5%	23.1%
Minimum	0.0%	0.0%	0.0%	0.0%
Maximum	46.2%	53.9%	53.9%	61.5%
P
Consultant	0.003	0.02	0.02	0.35
Experience	0.03	0.06	0.28	0.14
Number of Pythium cases seen	0.21	0.61	0.81	0.91

SD=standard deviation

Table 3 shows the identification rates by trainee status. Fellows identified a higher proportion of dot-like infiltrates (23% vs. 11.5%) and tentacles (35% vs. 23%) than consultants. Consultants identified a higher proportion of peripheral guttering (31% vs. 19%) than fellows. There was no effect of trainee status on the identification proportion of elevated plaques.

Table 3.

Summary of the Pythium identification rates by dot-like infiltrates, peripheral guttering, and tentacles, stratified by trainee status

Features	Summary	Fellow	Consultant	P
Dot-like infiltrates	Mean±SD	28.0%±14.5%	14.6%±11.9%	0.009
	Median	23.0%	11.5%
	25th percentile	23.0%	8.0%
	75th percentile	46.0%	23.0%
	Minimum	8.0%	0.0%
	Maximum	46.0%	38.0%
Peripheral guttering	Mean±SD	19.1%±13.3%	29.0%±12.2%	0.02
	Median	19.0%	31%
	25th percentile	8.0%	23.0%
	75th percentile	23.0%	38.0%
	Minimum	0.0%	0.0%
	Maximum	46.0%	54.0%
Tentacles	Mean±SD	33.4%±15.0%	22.2%±13.5%	0.01
	Median	34.5%	23.0%
	25th percentile	23.0%	15.0%
	75th percentile	46.0%	23.0%
	Minimum	0.0%	0.0%
	Maximum	54.0%	54.0%

SD=standard deviation

Discussion

Pythium keratitis is an emerging cause of visual morbidity mainly in tropical and subtropical regions.[8-11] Despite its wider geographic distribution, Pythium keratitis is still highly underdiagnosed and underreported, mainly because of a lack of awareness among clinicians regarding its specific clinical features and the nonavailability of standardized methods for accurate microbiological identification of the organism. In our study, the sensitivity and specificity in differentiating between Pythium and fungus by clinicians were 56% and 65%, respectively. Although the clinicians who participated in the study were primarily cornea specialists with years of experience, nearly half of them (47%) had seen less than five cases of Pythium keratitis per year. Three of the consultants with >10 years of experience have never diagnosed even a single case of Pythium in their career. We would attribute this primarily to the rarity of Pythium keratitis as well as lack of awareness among clinicians regarding this recent etiological agent and its specific features. There was no statistically significant difference in the differentiating ability between the consultants and fellows group.

An analogous study by Dalmon et al.[12] for assessing the diagnosing ability among clinical fellows and consultants, using clinical slit-lamp pictures for clinical differentiation of fungal and bacterial keratitis asserted that cornea specialists correctly differentiated more often than chance, but in less than 70% of cases. They were also able to specify the species, genus, and Gram stain accurately in 10%, 25%, and 46% of the time, respectively. In their study, presence of feathery border was strongly associated with fungal keratitis (sign was present in 14/39 photographs; P value 0.002, logistic regression), whereas bacterial keratitis was associated with a wreath infiltrate (seen in 3/30 photographs with a P value 0.005) or an epithelial plaque (P-value 0.02). But the main drawback of the study was the small number of raters (15 raters). Also, the study was conducted in two different countries, so there can be a lot of difference in the geographic prevalence of the common organism and its associated features.

Another study by Dahlgren et al.[13] used a similar survey method to evaluate the diagnosing ability of clinicians to predict the laboratory results of presumed microbial keratitis. Fifteen clinicians participated in the survey during a live examination of patients with suspected keratitis. The clinicians could successfully identify, among fungal, bacterial, and amebic keratitis, 73% of cases. The participants could correctly identify 79% of bacterial keratitis cases, 38% of fungal keratitis, and, 89% of amebic keratitis cases. However, this study included patients who came for primary care at the tertiary center and also those from referral clinics, thus there can be high chances of bias if referrals were deferred until distinctive clinical features were observed. Unfortunately, participants in this study, as well as those in a study by Sun et al.,[14] were not definite in diagnosing cases by uncommon organisms or those without the classic presentation. Thus, the study asserts that, clinicians were better in predicting the underlying cause with more common organisms (i.e., bacterial rather than fungal) or with organisms that demonstrate diacritic clinical features. Deep learning system demonstrated an overall accuracy of 76% for differentiating bacterial and fungal corneal ulcers from 1010 external photographs obtained by handheld digital camera.[15]

In our study, the proportion of correctly identified Pythium keratitis was less than that was expected. As geography influences the prevalence of each microorganism, the clinicians may not have equivalent clinical experience.

Among the various clinical manifestations of Pythium keratitis, the most distinguishing features include the presence of central/paracentral dense gray-white infiltrate with hyphate edges and flocculent debris overlying. Multiple mid-stromal “tentacle-like infiltrates,” “subepithelial expanding dot-like infiltrates,” “classical subepithelial or superficial stromal infiltrates radiating in a reticular pattern,” “peripheral guttering,” and “presence of raised epithelial plaques” have been described.[8,9,11] Radial keratoneuritis can be present. Associated stromal edema and Descemet’s folds may also be present. Hypopyon and endothelial plaques may accompany severe ulcers. Advanced cases may have corneal melts and perforation. Prime importance in prompt diagnosis lies in the fact that aggressive progression of Pythium keratitis may lead to corneal perforation, anterior chamber, limbal and scleral invasions, or endophthalmitis in a few days, leading to unsalvagable globe.[10,16] Pythium keratitis has high evisceration rates ranging from 53.3% to 91% in Thailand.[17-19] A recent study by Puangsricharern et al.[17] concluded that increased risk of globe removal is associated with the factors like delayed diagnosis, delayed initial treatment, increasing age, advanced disease, and dense hyphate edges.

Pythium antigen immunotherapy (PIAI) has also been studied, but its efficacy remains controversial.[17-19] Currently, therapeutic penetrating keratoplasty remains the most effective treatment, but it has to be done as early as possible. Nonetheless, recurrence of the infection and graft failure are very high, necessitating regrafting.

In spite of a large list of clinical signs available, clinicians consistently marked only a small subset of signs as most important in predicting the underlying etiology. The most helpful clinical sign as quoted by most clinicians (>50% of graders) in distinguishing Pythium keratitis included the presence of tentacles and peripheral guttering. A difficulty faced during the study was establishing the true presence of a clinical sign in a clinical photograph. Because of the lack of objective measurement, we defined a specific sign as present when >50% of raters were in agreement that it was depicted in the photograph. Overall, this study suggests the importance of being aware of specific clinical signs of Pythium keratitis and obtaining appropriate microbiological testing during the initial clinical suspicion itself, since this etiological organism could elude the treating ophthalmologists and be overlooked.

The main limitation of the study was that the clinical photographs were not able to completely capture the entirety of a slit-lamp biomicroscopic examination along with a patient history. A dynamic and 3D view, using different illumination techniques, would have been more informative. Although there were limitations, a picture-based survey provides unique advantages. Many physicians could identify the same keratitis individually in a masked manner and without much variability in the techniques of examination.

Conclusion

In conclusion, Pythium keratitis is often misdiagnosed as typical mycotic keratitis because of its morphological similarity. Clinical knowledge regarding the signs and symptoms and the microbiologically identifying features of Pythium keratitis are still lacking among clinicians. Since the treatments of fungal and Pythium keratitis are clearly different and individual organisms may be best treated by tailored therapy, it is very important to know the appropriate microbiologic tests and specific treatment based on the clinical picture during the initial clinical encounter itself.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.