Diagnostic value of microscopic blood smear examination using a digital cell morphology system for human granulocytic anaplasmosis

Abstract

Background

Human granulocytic anaplasmosis (HGA) is an emerging tick-borne zoonosis caused by Anaplasma phagocytophilum. HGA typically presents with flu-like symptoms and, in rare cases, can lead to life-threatening complications. Diagnosis relies on the polymerase chain reaction (PCR), which is the most sensitive, as well as serology or blood smear (BS) examination. However, PCR and serology are generally restricted to specialized laboratories and, depending on the organization and specialization of each laboratory, are often unsuitable for urgent diagnoses. Microscopic examination of BS can detect HGA by identifying intragranulocyte morulae but is time-consuming, requires trained microbiologists, and has limited sensitivity. This study evaluated the feasibility of morulae detection by BS analyzed with the CellaVision® system, a digital cell morphology platform for the automated classification of white blood cells and red blood cells, in patients with PCR-confirmed HGA.

Methods

Between 2019 and 2025, for each HGA PCR request at the Colmar Civil Hospitals, two May-Grünwald Giemsa-stained BS were prepared, digitized, and the cells preclassified using the CellaVision® System systematically. Retrospectively, two microbiologists performed a review of the recorded images for all positive HGA PCR results to identify intragranulocyte morulae independently and blinded to each other's findings.

Results

A total of 101 PCR tests for HGA were performed in 92 patients, with 5 patients testing positive (median time to results: 6 days). In all 5 patients, images of intragranulocyte morulae were obtained using the CellaVision® system, and the examination was completed in under 15 min.

Conclusions

Using the CellaVision® system for BS analysis with a standardized review protocol simplifies the reading process. The method successfully detected morulae in all confirmed cases included in this preliminary study though prospective validation is still required. However, a negative digitally assessed blood smear must not be interpreted as excluding HGA.

1. Introduction

Human granulocytic anaplasmosis (HGA) is a tick-borne zoonosis transmitted by Ixodes species (Acosta-Espana et al., 2025) and caused by Anaplasma phagocytophilum, an obligate Gram-negative intracellular bacterium that invades granulocytes. The first human cases were identified in the United States in 1994 (Chen et al., 1994), but HGA is now reported worldwide, with the highest incidence in North America (Kositz et al., 2025). The first French case was diagnosed in 2003 in Alsace (Northeast France) (Remy et al., 2003). A recent retrospective study identified 39 cases in Alsace between 2012 and 2024 (Gerber et al., 2025). HGA typically causes nonspecific flu-like symptoms and, rarely, life-threatening manifestations, making diagnosis challenging.

Diagnosis of HGA is based mainly on the polymerase chain reaction (PCR), which is the most sensitive technique (Hansmann et al., 2019). Serology and blood smear (BS) examination can also be performed. However, PCR and serology are generally restricted to specialized laboratories and are unsuitable for urgent diagnoses. Microscopic examination of BSs is easily accessible and can detect HGA by identifying intragranulocyte morulae. However, it is time-consuming, requires trained microbiologists, and has low sensitivity (20%–30%) (Hansmann et al., 2019). Digital morphology uses AI-driven algorithms to analyze and simplify BS, providing consistent and precise results. The CellaVision® DM1200 is such a digital cell morphology system for the automated classification of white blood cells and red blood cells (Panozzo et al., 2025). However; this system is used for an automated preclassification of with cells (granulocytes, monocytes, lymphocytes…) and an expert human validation is warranted to detect among preclassified granulocytes images of morulae. Such systems have been successfully used for malaria to identify images evocative of Plasmodium within red cells (Florin et al., 2018). The situation is quite similar for Anaplasma, as biologists must identify images suggestive of the bacterium (morulae) within the granulocyte population. However, examining conventional blood smears is tedious, given that neutrophil counts are generally low in cases of HGA and the proportion of neutrophils containing morulae is small (Hansmann et al., 2019). Consequently, pre-classification of blood smears using the CellaVision system could improve the identification of these cells. In this study, we evaluated the feasibility of microscopic examination performed using the CellaVision® DM1200 in patients with PCR-confirmed HGA.

2. Methods

Between October 2019 and June 2025, 101 PCR tests for suspected HGA were prescribed at the Colmar Civil Hospitals. For each request, two May-Grünwald Giemsa-stained BS were systematically digitalized on the CellaVision® DM1200 (Peripheral Blood Application), with the acquisition of 600 leukocytes per smear, totaling 1200 leukocytes per request. The CellaVision® system's automated analysis enables cell preclassification and image digitization.

During 2019–2022, the PCR techniques included a simplex real-time PCR test targeting the msp2/p44 gene of A. phagocytophilum (Koebel et al., 2012) (limit of detection: 40 bacteria/200 μL). Since 2022, a biplex PCR for simultaneous detection of Neoehrlichia mikurensis (Jahfari et al., 2012; Martinot et al., 2024) and A. phagocytophilum, in which the A. phagocytophilum target is still the msp2 gene, was used (limit of detection: 7 copies/μL). Serological testing was also ordered for some patients: Anaplasma phagocytophilum IFA (Test Kit “Focus Diagnostics”) with a screening cutoff of 1:64 for IgG and 1:20 for IgM.

For each request with a positive PCR result for HGA, or Neoehrlichia mikurensis, granulocyte images obtained with the CellaVision® system were manually reviewed retrospectively by two microbiologists, independently and blinded to each other's findings, using the CellaVision® software (Peripheral Blood Application), to identify intragranulocyte morulae. One typical image was considered sufficient to make the diagnosis. The microbiologists were aware of the positive PCR result, introducing a potential interpretation bias. The conventionally recognized morphological criteria were considered: intracytoplasmic in granulocytes, clusters of purplish basophilic elements and sharp outlines. These criteria were used for ruling out differential diagnoses (Dohle bodies, toxic granulations, and green granules of death).

For patient results, only descriptive statistics were performed (median and range). The BS detection proportion among PCR-confirmed cases with an IC95 was determined using the Clopper-Perason Method.

3. Results

3.1. Patients

A total of 101 PCR tests for HGA were performed in 92 patients, of whom 5 patients tested positive for HGA (median time to results: 6 days. Range, 3–9 days) and 1 for Neoehrlichia mikurensis (Fig. 1). Serologies were performed in 4 patients (median time to results: 8 days range 6–16 days) and 2 tested positive. For patients with HGA, clinical and laboratory results at admission are presented in Table 1. The median time from symptom onset to sample collection for BS preparation was 3 days (range, 1–11 days).

Fig. 1.

flow Chart of CellaVision retrospective analysis.

Table 1.

HGA Patients' Sociodemographic, clinical characteristics, and laboratory results upon admission (numerical values presented as median and range).

Age	72 years (range 52–87 y)
Sex	4 M/1 W
Reason for hospitalization	Fever (4 cases) Discomfort / Fall (1 case)
History of recent tick bites	3 cases
Duration of symptoms before hospitalization	3 days (range, 0–9 days)
C-reactive protein, mg/L	182 (120–247)
Leukocytes/mm3	5.310 (2.31–8.02)
Granulocytes/mm3	4.350 (1.110–6.740)
Platelets/mm3	52,000 (16000–233,000)
AST, UI/L	73 (25–236)
ALT, UI/L	43 (32–207)
LDH, UI/L	278 (224–493)
Treatment doxycycline first-line regimen doxycycline following beta-lactam spontaneous recovery	2 cases 2 cases 1 case
Outcome cure without sequalae	5 cases

3.2. CellaVision® DM1200 (Peripheral blood application)

In all five PCR-confirmed patients, the standardized lectures of the BS images using the CellaVision® system, with examination of preclassified granulocytes retrieved for all patients' images of intragranulocyte morulae (100% patients detected IC 95 47.8%–100%). The microbiologist agreement was 100% for the positivity of BS. The number of infected morulae was determined by consensus between the two microbiologists. The characteristics of each diagnostic sample are specified in Table 2. Patient 5 was hospitalized in the intensive care unit for 5 days due to acute kidney injury, rhabdomyolysis, and hemophagocytosis.

Table 2.

Characteristics of diagnostic samples (BS).

	Patient n°1	Patient n°2	Patient n°3	Patient n°4	Patient n°5
Leukocyte count (G/L)	3.35	8.02	7.40	6.30	7.08
Percentage of granulocytes	53.70	68.60	91.10	8	88
Granulocyte count (G/L)	1.8	5.5	6.74	0.5	6.25
Granulocytes infected count (1200 leukocytes)	2	7	21	1	21
Percentage of infected granulocytes	0.31	0.85	1.92	1.04	1.98
Active treatment against Anaplasma phagocytophilum before diagnostic sampling	No	No	No	Doxycycline (24 h)	No
Time between the onset of symptoms and collection of diagnostic sample (days)	11	9	1	6	5
Time between admission and collection of diagnostic sample (days)	8	0	0	1	2

Acquisition and automated re-classification of cell takes about 20–30 min but does not require human intervention. Lecture (i.e. the manual interpretation step) was completed in less than 15 min per patient. Morulae were clearly visible in granulocytes in all cases (Fig. 2).

Fig. 2.

Peripheral blood smears from five patients screened using the CellaVision DM1200 system, showing granulocytes containing intracytoplasmic morulae.

A 55 year-old woman was tested positive for Neoehrlichia mikurensis during the study period. Xamination of the leukocytes did not reveal any intracytoplasmic inclusions.

4. Discussion

HGA is an emerging tick-borne disease in numerous countries, including France. Anaplasma causes nonspecific symptoms (Acosta-Espana et al., 2025), and clinical diagnosis remains challenging, as other tick-borne diseases—such as neoehrlichiosis, babesiosis, tick-borne encephalitis, tularemia and rickettsiae—can present similarly. Although HGA often resolves spontaneously, it can occasionally lead to life-threatening complications such as acute hepatitis, renal failure, cytopenia, and hemophagocytosis (Dumic et al., 2022). Rapid diagnosis is therefore mandatory (Acosta-Espana et al., 2025), as mortality can reach 4%–6% (Dumic et al., 2022). Risk factors for severe disease include older age, hospitalization, and immunosuppression (Dumic et al., 2022). The characteristic laboratory triad of leukopenia, thrombocytopenia, and mild transaminitis is present in only about 25% of cases (Katragadda et al., 2025) and is not specific, as it can occur in other diseases, including babesiosis (Paleau et al., 2020). Current diagnostic tools, especially PCR, the most sensitive technique (Hansmann et al., 2019), are not universally available and often not performed routinely. In this study, the median delay to obtain the PCR was 6 days (delay between sampling and results). Serology shares similar limitations, with lower sensitivity due to delayed immunologic response (Hansmann et al., 2019; Ngo et al., 2025). Our laboratory does not perform PCR or serology testing for Anaplasma; these samples must be sent to more specialized laboratories, resulting in additional delays. Consequently, the time required to obtain results also depends on the organization and specialization of each laboratory. PCR Microscopic examination of peripheral BSs for intracytoplasmic inclusions within granulocytes (termed morulae) is specific but has low sensitivity; thus, their absence should not rule out the diagnosis (Ngo et al., 2025). However, BS examination is the only technique available in most hospital laboratories, although it is usually not performed, especially in high-income countries, due to the need for skilled biologists. It is time-consuming and has insufficient sensitivity (Hansmann et al., 2019) compared with PCR. However, BS results are available 24 h a day, 7 days a week, in most hospital laboratories. There is, therefore, a need to improve BS performance, as urgent treatment is critical and delayed treatment may lead to poor outcomes (Dumic et al., 2022). Moreover, doxycycline is used as first-line treatment in only about one-third of patients (Dumic et al., 2022), while beta-lactams—ineffective against HGA—are often prescribed by primary care physicians for fever even, including in our experience, after tick bites (Dumic et al., 2022; Martinot et al., 2025).

Digital cell morphology systems, used for the automated classification of white and red blood cells, are routinely applied in hematology to identify abnormal leukocytes and erythrocytes. Therefore, it is logical to extend their use to detect abnormal leukocytes due to non-hematological and infectious diseases (Panozzo et al., 2025; Florin et al., 2018). In HGA, some granulocytes contain morulae due to Anaplasma infection. The small number of affected granulocytes, combined with limited biologist training and lack of standardized BS reading, contributes to poor sensitivity and wide disparity in reported results observed in the literature. However, by classifying and concentrating granulocytes, our study showed that digital cell morphology systems, with a standardized method, could be of great help for biologists and highlights the feasibility of the technique for Anaplasma BS analysis. Rand et al., in a series of 14 confirmed HGA cases, reported 100% sensitivity while examining 200 granulocytes (Rand et al., 2014). Another author suggests that morulae may be present in less than 0.1% of neutrophils (Thomas et al., 2009). In our series, the examination of 1200 leukocytes allowed the detection of morulae in all cases though it may require an additional BS analysis, particularly in cases of profound neutropenia. Although human verification remains necessary in this study, future approaches could incorporate fully automated processes using AI, potentially improving both sensitivity and processing times. Lastly, we can make the hypotheses that apart from the initial cost of the digital cell morphology system, the analysis itself is inexpensive, as it does not require reagents and reduces the time required by laboratory personnel.

Our study has limitations. First, it is a retrospective study in which biologists were not blinded to the results of the PCR for this reading. The sensitivity could therefore be artificially high. Microbiologist should exercise a degree of caution especially in the case of single positive cells. Second, the extremely small number of positive cases (n = 5) precludes a robust estimation of diagnostic performance and significantly limits the generalizability of the results. Third, specificity could not be assessed, as no non-HGA pathologies (except for one neoehrlichiosis) were included, and morphologically similar intracytoplasmic inclusions may be observed in other conditions, such as green neutrophilic inclusions or Döhle bodies. Therefore, this study appears to be an interesting ‘proof of concept’ in favor of digital cell morphology systems for the diagnosis of HGA. Only a prospective study could, however, corroborate these results and assess the true sensitivity and specificity of the technique. Notably, the inability to estimate specificity constitutes another significant limitation of the study; the inclusion of patients who test negative by PCR but exhibit similar hematological abnormalities in prospective studies will be necessary before clinical implementation can be recommended. We did not include a comparison group for the BS. examinations performed using conventional microscopy; therefore, this study does not allow for an assessment of whether CellaVision analysis was more sensitive, faster, or easier to implement than conventional microscopy. Finally, not all laboratories are equipped with digital cell morphology systems, but their use is likely to increase for hematological diseases, and our study highlights their potential application for HGA. Digital cell morphology system could also in the future be used, provided adapted tools are designed, at the same time to screen babesiosis, which also produces pathological images in infected red blood cells with a very low number of infected cells, especially in immunocompetent patients, leading to a very difficult diagnosis (Martinot et al., 2011). A modification to the PCR technique occurred during the study period; however, the techniques targeted the same gene and were developed by the same laboratory (Strasbourg University Hospitals). Consequently, they did not alter, in our view, the detection proportion among PCR-confirmed cases or the specificity of the results, and thus had no impact on our conclusions. As our results were based exclusively on positive PCR tests, this change had no impact on our observations. Neoehrlichia mikurensis is an emerging tick-borne pathogen in France (Martinot et al., 2024; Boyer et al., 2021). In our case of Neoehrlichia mikurensis infection, microscopic examination did not reveal any morulae. The diagnosis of these infections relies on specific PCR; to our knowledge, there is no published data demonstrating the presence of morulae in leukocytes in neoehrlichiosis. However, among other TBD, Ehrlichiosis, a bacterial infection that is exceptional in Europe, is known to possibly result in morulae images in granulocytes or macrophages. Thus, we could make the hypothesis that the digital morphology system could also be for Ehrlichiosis a possible tool for this disease (Thomas et al., 2009; Emiroglu and Celebi, 2019).

In conclusion, this study using the CellaVision® digital system successfully detected morulae in all PCR-confirmed cases of HGA and warrants prospective validation in larger studies on diagnostic accuracy. A negative digitally assessed blood smear must not be interpreted as excluding HGA.

IA Statement

Generative AI or AI-assisted technologies were not used in the preparation of the manuscript.

CRediT authorship contribution statement

Olivier Augereau: Writing – review & editing, Writing – original draft, Validation, Methodology, Investigation, Formal analysis, Conceptualization. Guillaume Gregorowicz: Writing – review & editing, Validation, Resources, Methodology, Investigation, Conceptualization. Victor Gerber: Writing – review & editing, Validation, Methodology, Investigation, Conceptualization. Thomas Gueudet: Writing – review & editing, Investigation. Loup Lang: Writing – review & editing, Investigation. Dominique De Briel: Writing – review & editing, Validation, Resources, Methodology, Investigation, Conceptualization. Martin Martinot: Writing – review & editing, Writing – original draft, Validation, Methodology, Investigation, Formal analysis, Conceptualization.

Consent for publication

NA

Ethics approval and consent to participate

This retrospective study was approved by the Ethics Committee of Medicine Odontology and Pharmacy Faculties and Hospitals [University Hospital of Strasbourg; N° CE-2025-269). In accordance with French regulations, participation was based on the absence of expressed opposition after receipt of the study information.

Funding

None.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Data supporting the conclusions of this article are included in the article.