Hypereosinophilia: Biological investigations and etiologies in a French metropolitan university hospital, and proposed approach for diagnostic evaluation

Objectives We aimed to evaluate the usefulness of biological investigations in cases of eosinophilia in our area (French Alps). Methods We retrospectively included all adult patients attending the infectious disease and internal medicine units between 2009 and 2015 with eosinophilia ≥1 G/l. Results We identified 298 cases (129 women and 169 men). In 139 patients, eosinophilia had not been addressed. In the 159 others, the cause of eosinophilia was identified in 118 (74.2%). The main identified causes at the time were drug reactions (24.5%, mostly β-lactams and allopurinol), infectious diseases (17.0%), vasculitis (8.2%), autoimmune diseases (6.9%), and malignant diseases (6.2%). In patients with a skin rash, eosinophilia was significantly more often investigated, and a diagnosis significantly more often made. Helminthosis were mainly diagnosed in tropical travelers (18/24) excepting toxocariasis (3 non-travelers). Stool examination for helminthosis was positive in 5/76 patients (6.6%) (all tropical travelers); 391 helminth serologies were performed in 91 patients, with 7.9% being positive (all but 3 positive cases were travelers). Anti-neutrophil cytoplasmic antibodies (ANCA) were positive in 26/112 patients (23.2%), with 9 cases of vasculitis identified. Conclusions Drug-related eosinophilia is the main etiology. Search for helminthosis is not recommended among non-travelers (excepting toxocariasis). ANCA should be performed early so as not to overlook vasculitis.

Introduction Blood eosinophilia is defined as a level of eosinophilic granulocytes above 0.4 or 0.5 G/l (1 G/ l = 10 9 cells per liter) on the blood count; the levels of blood eosinophilia may be described as mild (0.5-1.5 G/l), moderate (1.5-5 G/l), or severe (>5 G/l) [1]. Eosinophils are key effectors of innate immunity against helminth parasites and as part of the allergic inflammation. After being produced in bone marrow, they circulate for only a few hours before being recruited in tissue where they act in diverse ways: degranulation (sudden release of the highly reactive content of their granules), cytokine production, and phagocytosis; blood eosinophilia is not strictly correlated with tissue infiltration by these cells. The incidence of eosinophilia can dramatically change depending on the world region or the existence of a recent travel history. In large transversal studies of routine medical samples, blood eosinophilia has been estimated to affect 0.4 to 4% of all blood counts [2,3], whereas it has been found in 4 to 27% of returning travelers or arriving refugees [4,5]. When eosinophilia is present, the suspected diagnoses may vary depending on the conditions: for example, in returning travelers, 18.9 to 53.7% of eosinophilia cases are related to an helminth infection [4,6]. However, eosinophilia is associated with a broad variety of non-helminth diseases (including hematological malignancy, vasculitis, allergic diseases, and hypereosinophilic syndrome). Because of this diversity, the diagnostic approach may become complex. Several authors have reviewed the diagnoses associated with blood eosinophilia [7][8][9][10][11], although the diagnostic strategy, which is strongly dependent on the clinical condition and circumstances, is less often detailed. Indeed, recommended examinations may differ from one publication to another.
We therefore aimed to detail in a retrospective study the etiologies of eosinophilia in Grenoble University Hospital (French Alps) and determine which laboratory tests are profitable in this context.

Materials and methods
All patients with an eosinophilia !1 G/L who attended the internal medicine or infectious disease units of Grenoble University Hospital between 2009 and 2015 were included. The subjects may have consulted or been hospitalized. We choose this threshold (1 G/L) by assuming that a high proportion of cases of eosinophilia between 0.5 and 1 G/L might have been overlooked by clinicians.
Clinical and biological data were collected. All elevated values of aspartate transaminase and/or alanine transaminase (i.e., at least twice the upper normal value) during the period of eosinophilia were considered to be "cytolysis." Elevated alkaline phosphatase and/or gamma glutamyl-transferase (same) were considered to be "cholestasis." The diagnosis confirmed at the time by the medical team was retrieved from the medical records.
Chi-square and Fischer tests were used for intergroup comparisons of categorical variables for groups of more or less than 20 patients, respectively. Student's t-test was used for quantitative variables. P-values <0.05 were considered to be statistically significant.

Eosinophil count and population
The population was composed of 129 women and 169 men, with a median age of 66 [25 th percentile = 47; 75 th percentile = 80; range 13-99] (Fig 1). The mean maximal eosinophilia was 2.46±1.7 G/L (median 1.6 G/L; 25 th percentile = 1.1; 75 th percentile = 2.3). There was no significant difference between men and women with respect to their age and value of their maximal eosinophilia.

Etiologies of eosinophilia
Etiologies of eosinophilia are summarized in Table 1. The existence of eosinophilia was not taken into account in 139 (46.6%) of the 298 selected cases. This proportion was higher for mild eosinophilias (92 of 153 cases [60.1%] with an eosinophilia <1.5 G/l; 46 of 120 cases [38.3%] with an eosinophilia between 1.5 and 5 G/l; and 1 of 25 cases [4.0%] with an eosinophilia >5 G/l, p<0.01). The cause of eosinophilia was identified in 118 cases (74.2% of the 159 explored cases, and 39.6% of the entire population). In 41 patients, the eosinophilia was explored or mentioned in the medical records, but without diagnosis.
Drug-related eosinophilia. The most frequently identified cause of eosinophilia was a drug reaction: 39 patients (13.1% of the entire population, and 24.5% of the population whose eosinophilia was taken into account) received this diagnosis, including 10 with a diagnosis of drug reaction with eosinophilia and systemic symptoms (DRESS); 60.0% of patients with DRESS had a fever. Allopurinol alone was involved in six cases of eosinophilia, including three DRESS. Antibiotics were also well represented: a beta-lactam-induced eosinophilia was identified for 14 patients (4.7% of the whole population) (piperacillin/tazobactam in four cases and amoxicillin in three cases), while trimethoprim/sulfamethoxazole was involved in three cases. Incriminated molecules are listed in Table 2. For two patients, a drug reaction was identified despite no causative agent being identified. Globally, patients with drug-related eosinophilia were significantly older (mean age 70.4±8 vs 61.1±10, p<0.05).
Infectious diseases. A diagnosis of infection was observed for 17.0% (27 of the 159 patients whose eosinophilia was taken into account). Among the 24 cases of helminth infection as the source of eosinophilia (Table 3), 17 were biologically proven, whereas it was only suspected for seven on the basis of a suggestive clinical context (including four tropical/ subtropical travelers) and the success of empirical treatment. Identified parasites were mostly tropical (including four with schistosomiasis), although three Toxocara canis infections were found. A non-helminth infection was identified in three cases.
Others. Eosinophilia was attributed to vasculitis for 13 patients (4.4% of the whole population) and autoimmune disease for 11 patients (3.7%). A hypereosinophilic syndrome was diagnosed in five cases and a solid tumor IL-5 expression in five cases. An isolated organ eosinophilic infiltration was detected in six patients.

Severe eosinophilia
Overall, 25 patients had a maximal eosinophilia !5 G/l, of which 24 cases were explored. The examinations performed led to the diagnoses of infectious disease (three cases), hematological malignancy (four cases), and vasculitis (four cases).
Travel. Among the 40 tropical travelers, helminth infections were the most commonly identified cause of eosinophilia (18 cases, 45.0% of travelers). The period between the travel and the first reported eosinophilia was less or more than a year for 21 patients and eight patients, respectively, and unknown for 11. Destinations where helminth infections were acquired are listed in Table 3. Skin rash. Overall, 78 patients had a skin rash during eosinophilia (26.2% of the whole population [78 out of 298]). When it was accompanied by a skin rash, the eosinophilia was significantly more often taken into account (63 out of 78 patients with skin rash [80.8%] vs 96 out  (Fig 3). Fever. A total of 141 patients presented with fever. Notably, 56.4% of drug reactions (including 60.0% of DRESS cases) had a body temperature >38.5˚C. When explored, the cases  Investigation of hypereosinophilia with both eosinophilia and fever received a diagnosis of drug reaction in 29.7% of patients, and a diagnosis of infectious diseases in 10.8%.
Digestive symptoms. Digestive symptoms were observed in 70 patients; 11 had a diagnosis of helminth infection, of whom 9 had a recent travel history.

Laboratory tests
Stool examination. A stool examination was performed in 76 cases (25.5%). One or two (negative) stool sample(s) were collected from 46 patients, i.e., under the recommended number of three stool examinations. For 25 other patients, this examination was performed three times without identifying a parasite. For five more patients, a stool examination was positive; all five patients had returned from a tropical country. Among the 222 patients who did not have a stool examination, three received a diagnosis of helminthosis, with one case each of schistosomiasis, onchocerciasis, and toxocariasis.
Serologies. A total of 391 helminth serologies was performed in 91 patients (including 173 serologies in 30 travelers). Toxocara serology was the most frequently used (72 times). Overall, 31 serologies were positive in 20 patients (seven had two or more positive serologies). Among the 31 positive results, 11 (35.4%) led to a diagnosis of parasitosis in 11 cases (eight travelers and three cases of toxocariasis). For two other patients, parasitosis was detected by other means.
Immunoglobulin E. Total immunoglobulin E (IgE) was determined in 51 patients. Elevated IgE levels (>150 UI/ml) were found in 24 cases. This condition was not specific, as it was associated with helminth infection (six cases), allergic diseases (four cases), and vasculitis (five cases).

Discussion
In our center, etiologies associated with an eosinophilia were dominated by drug reactions and helminth diseases. Strikingly, 46.6% of the 298 eosinophilia cases were probably neglected (and not even mentioned in the medical records) despite the fact that the level of eosinophils chosen for the selection of patients was high (!1 G/l). A higher level of eosinophils was significantly more often taken into account. Thus, although the normal value of eosinophils is less than 0.4 or 0.6 G/L, this would suggest that most physicians do not pay attention to a transitory mild eosinophilia, unless it is associated with clinical or non-clinical disorders. The proportion of neglected cases would probably have been higher if we had retained a lower threshold value.
The most common cause observed was drug-related (24.5% of investigated eosinophilias), and the most frequently incriminated drug was allopurinol. Many other reactions (including DRESS) were related to antibiotics. These medications could have been over-represented as patients were selected (in part) from an infectious diseases unit. However, our results correspond to the international RegiSCAR study on DRESS [12] in which a high proportion of the 117 cases involved antibiotics and allopurinol. Noticeably, fever accompanied 90% of DRESS cases in the RegiSCAR study compared to only 60% in our own. This difference may be explained by the threshold value choosen to define fever (38˚C for Kardaun et al. and 38.5˚C for our study). In our study, drug-related eosinophilia was also associated with a significantly older age, which may be explained by the higher number of medications prescribed to older patients. However, as every age group was concerned, age should not be used as a determining factor in the diagnostic process.
A skin rash was reported in 78 patients, with strong implications on the diagnostic process but in different ways. First, in the presence of a skin rash, eosinophilia was more likely to be taken into account, which could be related to the fact that the leukocyte differential received more attention when a skin rash was present. Second, the combination of eosinophilia and skin rash was strongly associated with identifying an iatrogenic cause. This may be a key element in the diagnostic approach, as it could avoid excessive lab evaluation for patients who present eosinophilia with skin manifestations.
Helminth infections were not infrequent, but mostly concerned returning travelers. The only helminth identified in non-travelers was Toxocara, an ubiquitous zoonotic roundworm whose larvae can accidentally penetrate the human body. Only five stool examinations (in 76 patients) permitted the identification of a parasite (or its eggs), and all of them concerned travelers; this should be nuanced by the fact that many stool examinations were not performed the recommended three times. Our results suggest that helminth serologies are cost-effective labs in travelers (except for toxocariasis), as they allow the diagnosis of parasitosis to be strengthened or confirmed; Schistosoma, Strongyloides, and Toxocara serologies were particularly useful in comparison to others. Thus, we confirmed that helminth infections, though not rare, Table 4. Recommended examinations to explore eosinophilia according to seven previous studies. "Yes": recommended first-line examination; "2 nd ": recommended second-line examination. should be preferentially searched for in travelers. We propose for stool examinations and helminth serologies to be reserved for patients returning from a tropical country and/or with suggestive symptoms of a given helminth disease, excepting the Toxocara serology, which may be relevant in asymptomatic patients and non-travelers. The Strongyloides serology might also be performed early in the diagnostic process, because this disease can be discovered years after the travel during which the infection was acquired; the existence of a recent travel is then irrelevant. Despite the fact that hypereosinophilia is not associated with this disease, alveolar echinococcosis was investigated in many patients (probably because it is known to be endemic in our region). This parasite was never found among the 298 cases of eosinophilia in our study. Vasculitis and other autoimmune diseases were not rare in this study, which may be due to the fact that patients were selected in part from an internal medicine department. However, this may also reflect the fact that these diagnoses are frequent in patients with eosinophilia >1 G/L; given the potentially severe complications of these diseases, anti-neutrophil cytoplasmic antibody testing should be performed early in these patients.

Recommended exams
It is noteworthy that apart from allergy (including drug reaction), helminth infection, and vasculitis and other autoimmune diseases, more rare diagnoses should be kept in mind, including blood malignancy (particularly Hodgkin disease), and hypereosinophilic syndrom. Persistent eosinophilia in non-travelers patients not receiving any therapy should not be left aside, and such rarer diagnoses should always be evoked; the last is an excusion diagnosis, and should be considered solely if other causes have been ruled out.
Even when eosinophilia was explored, many patients did not receive a diagnosis (25.7% of investigated cases), thus reflecting the difficulties encountered to identify the causes of eosinophilia. The complexity of the diagnostic process is illustrated by the heterogeneity of different recommendations proposed in literature [1,7,8,9,11,[13][14] (Table 4).
In previous studies, the frequencies of the diagnoses were different. In the study of Sade et al [15], on 100 patients with an eosinophilia above 0.650 G/l, drug reactions and helminth infections were less frequent (6% and 5%, respectively), whereas asthma and allergic diseases were more prevalent (13%); the proportion of unexplained eosinophilia and hypereosinophilic syndrome were higher than in our study (34% and 7%). In the study of Lombardi and Passalacqua [16], on 1862 patients with eosinophilia above 0.35 G/l, 79.7% of cases were related to asthma or allergy, 8.2% to helminth diseases, and 0.6% to autoimmune diseases. These major differences may be due to the low threshold of eosinophils chosen by the authors to select the patients, since asthma and allergy are more often correlated with mild eosinophilia. Strangely, no drug reaction were reported as cause of eosinophilia in this study.
Finally, we propose a strategy based on our results (Fig 4) that draws from skin changes, drug prescriptions, and recent travel history in tropical regions, as we observed that travel and drug reaction are prominent causes of eosinophilia. In case of lymph node enlargment, a diagnosis of neoplasm (prominently Hodgkin disease) should be considered in the first place.

Conclusion
Mild eosinophilia is often neglected. A diagnosis of drug-related reaction is frequently identified, particularly in case of rash. Apart from toxocariasis, helminth infections should be searched for (mostly by serology or stool examination) in travelers. ANCA should be performed early, as potentially severe vasculitis is not uncommon. Determination of serum IgE level is weakly informative.