Rapid antifungal susceptibility testing from positive blood cultures using direct gradient diffusion: a prospective study

Abstract

Objective

To evaluate the agreement between direct gradient diffusion (DGD), performed directly from positive blood culture bottles, and the Sensititre^™ YeastOne^™ (SYO) method for antifungal susceptibility testing in Candida spp.

Methods

A prospective observational study including blood cultures positive for yeasts between June and October 2024 was conducted. DGD (ETEST^®) was performed directly from blood culture broth and compared with SYO. The antifungal agents evaluated were anidulafungin, micafungin, caspofungin, fluconazole, voriconazole, and amphotericin B. Categorical agreement (CA) and essential agreement (EA) were analyzed.

Results

A total of 32 Candida isolates were included. Overall CA was 90%, whereas EA was 60.6%. CA was 100% for amphotericin B, 92.5% for azoles, and 86% for echinocandins. No very major errors were observed. However, all Candida parapsilosis isolates showed major errors for anidulafungin, appearing resistant by DGD and susceptible by SYO. All DGD results were interpretable after 24 hours.

Conclusions

DGD performed directly from positive blood culture bottles provides early antifungal susceptibility results with high CA and markedly reduced turnaround time. However, caution is required when interpreting anidulafungin resistance in C. parapsilosis. This approach may serve as an adjunctive tool for early optimization of antifungal therapy.

Introduction

Invasive fungal infections (IFIs) have increased substantially over recent decades, driven by the growing number of immunocompromised patients, the widespread use of broad-spectrum antibiotics, and the increasing use of invasive medical devices [1,2]. Although Candida albicans remains the most frequently isolated species, non-albicans Candida species have become increasingly prevalent and often display less predictable antifungal susceptibility patterns [2,3].

Broth microdilution, as standardized by CLSI and EUCAST, remains the reference method for antifungal susceptibility testing [4]. However, these methods are labor-intensive and time-consuming, limiting their value for early clinical decision-making. Sensititre YeastOne^® (SYO) (Thermo Fisher Scientific, USA) is a simplified and widely used commercial colorimetric method, although results are still generally available only 24–48 hours after isolate recovery [5–7].

The clinical need to shorten this interval has led to the development of rapid antimicrobial susceptibility testing (RAST) approaches for selected bacterial pathogens [8]. However, the application of similar strategies in clinical mycology remains limited. Extending this concept to antifungal susceptibility testing by applying direct gradient diffusion (DGD) to positive blood culture bottles could improve early antifungal stewardship and reduce the consequences of delayed effective therapy.

To our knowledge, few studies have evaluated DGD performed directly from positive blood culture bottles for antifungal susceptibility testing. The aim of the present study was to evaluate the agreement between DGD and SYO, and to determine whether this strategy could provide clinically useful antifungal susceptibility data earlier than the conventional workflow.

Material and methods

Study design and isolates

We performed a prospective observational study including patients with blood cultures positive for yeasts between June and October 2024. Blood culture samples were processed using the automated BD BACTEC^™ system (Becton Dickinson, Sparks, MD, USA) with BD BACTEC^™ Plus Aerobic/F culture vials.

Species identification was performed by MALDI-TOF MS (Bruker Daltonics, Germany) after 24 hours of incubation on Sabouraud agar at 37°C.

Antifungal susceptibility testing

DGD testing (E-test, bioMérieux, France) performed from positive blood culture bottles was compared with the SYO method. For gradient diffusion testing, 125 μL of blood culture broth was inoculated directly onto RPMI agar plates (bioMérieux, France) without prior measurement or standardization of inoculum density (e.g., via McFarland standard or quantitative colony counts). The antifungal agents tested included the echinocandins anidulafungin, micafungin, and caspofungin; the azoles fluconazole and voriconazole; and amphotericin B. MICs were read after 24 hours of incubation at 37°C. SYO testing was performed according to the manufacturer’s instructions. Susceptibility results were interpreted according to CLSI criteria.

Definitions

Categorical agreement (CA) was defined as the proportion of isolates yielding the same categorical interpretation by both methods. Essential agreement (EA) was defined as an MIC obtained by gradient diffusion falling within ±1 twofold dilution of the corresponding SYO MIC; an EA rate of ≥90% was considered acceptable. Discrepancies were classified as minor errors (mE) when one method classified an isolated as intermediate and the other method classified it as susceptible or resistant; major errors (ME) when the rapid method yielded false resistance; and very major errors (VME) when the rapid method yielded false susceptibility [9].

Statistical analysis

Agreement between methods was evaluated using CA and EA. Results were also analyzed according to antifungal class and species, with particular attention to clinically relevant discrepancies. CA, EA, and error rates were calculated with corresponding 95% confidence intervals (95% CIs) using the Wilson score method.

Results

A total of 32 Candida isolates were analyzed: C. glabrata (n = 14), C. parapsilosis (n = 7), C. albicans (n = 7), C. tropicalis (n = 2), C. dubliniensis (n = 1), and C. krusei (n = 1).

Overall, CA was 90%, whereas EA was 60.6%. When analyzed by antifungal class, amphotericin B showed the highest CA (100%), followed by azoles (92.5%) and echinocandins (86%). Amphotericin B also showed the highest EA (74.2%) (Table 1).

Table 1.

Categorical agreement and essential agreement between direct gradient diffusion and Sensititre YeastOne^® according to antifungal class

Antifungal class	N	CA, % (95% CI)	EA, % (95% CI)	mE, % (95% CI)	ME, % (95% CI)
Echinocandins	93	86.0 (77.5 – 91.7)	60.2 (49.9 – 69.7)	6.5 (2.9 – 13.4)	7.5 (3.7 – 14.7)
Azoles	46	92.5 (81.3 – 97.2)	60.0 (45.5 – 72.8)	4.4 (1.2 – 14.8)	2.2 (0.4 – 11.7)
Amphotericin b	31	100.0 (89.0 – 100.0)	74.2 (56.8 – 86.3)	0.0 (0.0 – 10.7)	0.0 (0.0 – 10.7)
Overall (excl. Anidulafungin)	139	95.0 (89.8 – 97.6)	64.0 (55.8 – 71.5)	4.3 (2.0 – 9.1)	0.7 (0.1 – 4.0)
Global	170	90.0 (84.6 – 93.7)	60.6 (53.1 – 67.6)	5.3 (2.8 – 9.8)	4.7 (2.4 – 9.0)

CA: categorical agreement; EA: essential agreement; mE: minor errors; ME: major errors.

No VME were observed in this series. However, a clinically relevant discrepancy was identified for C. parapsilosis: all isolates yielded ME for anidulafungin, appearing resistant by DGD and susceptible by SYO (Table 2).

Table 2.

Comparison of anidulafungin MICs obtained for C. parapsilosis isolates using E-test and Sensititre YeastOne^®

Isolate	E-test MIC (mg/L)	Interpretation	SYO MIC (mg/L)	Interpretation	Error
1	16	R	0.5	S	ME
2	32	R	0.5	S	ME
3	4	R	0.5	S	ME
4	>32	R	2	S	ME
5	>32	R	2	S	ME
6	12	R	2	S	ME
7	>32	R	0.5	S	ME

ME: major errors.

Importantly, all gradient diffusion tests were interpretable after 24 hours, providing results approximately 48 hours earlier than the conventional workflow.

Discussion

In this study, DGD performed directly from positive blood culture bottles showed high overall CA with SYO and substantially reduced turnaround time. These findings suggest that this approach may be a useful adjunctive tool for early antifungal therapeutic guidance in patients with candidemia.

The overall CA of 90% supports the potential clinical applicability of the method, particularly as an early reporting strategy. These results align with recent findings by Miaux et al. [10], who evaluated direct gradient diffusion against standard colony-based E-test from positive blood cultures in candidemia patients, also demonstrating high overall categorical agreement (>90%) across fluconazole, voriconazole, amphotericin B, and anidulafungin, alongside a turnaround time reduction of at least 24 hours.

By contrast, the overall EA was markedly lower (60.6%), indicating considerable variability in MIC values between the two methods. This discrepancy is not unexpected and likely reflects intrinsic methodological differences between agar-based DGD and broth-based microdilution techniques. In addition, trailing growth with azoles and echinocandins may complicate endpoint determination and increase interobserver variability, especially when the 80% inhibition threshold must be identified visually [11–13].

A particularly relevant finding was the consistent overestimation of anidulafungin resistance in C. parapsilosis by DGD. This observation is consistent with previous reports describing reduced agreement for this specific species-drug combination [14–16]. Although the underlying explanation remains uncertain, species-specific growth characteristics, antifungal diffusion dynamics in solid media, and methodological differences between agar-based and broth-based systems may contribute to this pattern. From a practical perspective, anidulafungin resistance results obtained by DGD in C. parapsilosis should be interpreted with caution and confirmed by a reference or otherwise validated method before being used to guide definitive therapy.

The clinical relevance of shortening susceptibility turnaround time is considerable. In candidemia, delays in the initiation of effective antifungal therapy have repeatedly been associated with worse outcomes, including increased mortality [17]. In this context, the possibility of obtaining an interpretable susceptibility profile within 24 hours of blood culture positivity may support earlier therapeutic optimization, whether in the form of treatment escalation, de-escalation, or refinement of empirical therapy. Even if rapid testing is not intended to replace standard susceptibility methods, it may still provide clinically meaningful early guidance.

This study has several limitations. First, the number of isolates was limited, and some species were underrepresented, thereby restricting the generalizability of the results and the robustness of subgroup analyses. Second, the study was performed at a single center, and local epidemiology or laboratory workflow may have influenced performance. Third, direct inoculation was performed using a fixed volume of broth (125 μL) without prior inoculum standardization. Although yeast concentrations in positive bottles (– CFU/mL) do not significantly alter gradient diffusion MICs, omitting cell density adjustment introduces inherent inoculum variability [18]. Fourth, the study did not include comparison with a reference broth microdilution method, such as CLSI or EUCAST, which precludes definitive attribution of discrepancies to either DGD or SYO. Fifth, clinical outcomes such as mortality, time to appropriate therapy, and treatment modification were not evaluated. Establishing a direct relationship between earlier in vitro susceptibility results and patient outcome in candidemia is inherently challenging because of heterogeneity in host factors, source control, illness severity, and concurrent management decisions.

Taken together, our findings support the potential role of DGD from positive blood culture bottles as an adjunctive early susceptibility strategy rather than as a replacement for standard methods. Its principal strength lies in the marked reduction in turnaround time, whereas its limitations relate mainly to MIC variability and the suboptimal performance observed for anidulafungin in C. parapsilosis. Future studies including larger series, reference broth microdilution methods, and clinical outcome assessment are warranted to confirm the utility of this approach.

Conclusions

DGD performed from positive blood culture bottles appears to be a rapid and operationally attractive approach for early antifungal susceptibility testing in Candida spp. Its implementation may allow preliminary susceptibility results to be reported 24–48 hours earlier than with conventional workflows based on methods such as Vitek^®2 or SYO, thereby supporting more timely clinical decision-making. Although this method appears to overestimate anidulafungin resistance in C. parapsilosis, the absence of VME and the high overall CA support its potential value as an early guidance tool. Accordingly, DGD may represent a useful adjunct for the early optimization of antifungal therapy in critically ill patients.

Funding

No funding was received for the research, authorship and/or publication of this article.

Conflicts of interest

I declare that there is no commercial or financial conflict of interest for this research.