3. Results
From January 1984 to January 2014 (30 years), 76 patients with febrile neutropenia related to non-chemotherapy drugs, at time of the discovery, were registered. At the time of detecting this drug-related adverse effect, eight patients (10.5%) were already hospitalized for another reason (“in patients”); the other patients were treated at home (“out patients”).
3.1. Patient Baseline Characteristics
All patients were Caucasian (n = 76). Mean and median age were 52.2 and 56 years (range: 18–93), respectively, with 16 patients (21.1%) younger than 50 years, and 51 (67.1%) younger than 75 years. Six patients were aged 85 years or more (7.9%). The female/male ratio was 1.6.
Several comorbidities were found in 86.8% of cases (n = 66), consisting primarily of: thyroid disorders (mainly Grave’s disease) (n = 17, 22.4%); arterial hypertension (n = 13, 17.1%); cardiac disorders (cardiac failure, atrial fibrillation, myocardial infarction, arteritis of lower limbs, or stroke) (n = 13, 17.1%); chronic renal failure (creatinin clearance <60 mL/min) (n = 13, 17.1%); neuro-psychiatric disorders (psychosis, schizophrenia, dementia, epilepsy, anorexia, and depression) (n = 11, 14.5%).
3.2. Causative Drugs
A single drug was documented as “causative” or “likely causative” in all except 12 cases (15.8%), for which two to four drugs were suspected. The causative drugs have been listed in
Table 2. The respective causative drugs were stopped within the first 48 h of admission in 123 patients (60.6%).
The main drug families found to be causative were: antibiotics (
n = 37, 37.4%), especially β-lactams (
n = 21) and cotrimoxazole (
n = 6); antithyroid drugs (
n = 17, 17.2%); neuroleptic and anti-epileptic agents (
n = 13, 13.1%); non-steroidal anti-inflammatory agents and analgesics (
n = 8, 8%); platelet aggregation inhibitors (
n = 8, 8%), especially ticlopidine (
n = 5) (
Table 2). Since 1990 and 2000, no case of noramidopyrine- and ticlopidine-induced agranulocytosis has been observed, respectively. Conversely, cases of febrile neutropenia associated with either clozapine or antiviral agents have been observed only since 2000 and 2005, respectively. With the exception of four, all cases of antibiotic-induced neutropenia were observed in hospitalized patients.
In half of cases (n = 40), patients were treated with at least three drugs (mean number of drugs: 5.2; range: 1–13). Overall, 21 patients (27.6%) did not take another drug in addition to the causative drug within the 10 days preceding febrile neutropenia detection. There was only one case of self-medication. The mean and median durations of the suspected drug intake were 54 and 30 days (range: 3–120), respectively. A quarter of the patients received the causative drug for at least 30 days.
3.3. Clinical Manifestations
The main discovery circumstance, based on the inclusion criteria used (febrile neutropenia), was fever occurring as isolated fever (of unknown origin) in 70 patients (92.1%). Six patients (7.9%) exhibited fever related to pre-existing pneumonia (n = 2), pelvic and abdominal abscess (n = 2), and meningitis (n = 2).
The main clinical presentations upon hospitalization were as follows: isolated fever (
n = 23, 30%); sore throat, acute tonsillitis and sinusitis (
n = 14, 18.4%), with one case of maxillary sinus mucormysosis; documented pneumonia (
n = 14, 18.4%), with one case of pulmonary aspergillosis and one case of acute respiratory distress syndrome; septicemia (
n = 11, 14.5%); septic shock (
n = 5, 6.6%).
Table 3 summarizes data of the remaining symptomatic patients with documented infection (
n = 9, 11.8%).
Microbiological documentation was obtained from 21 patients (31.8%) (data available for
n = 66), with mainly Gram-negative bacilli (
n = 8) and
Staphylococcus sp. (
n = 7) (
Table 4).
While in hospital, 22 patients (28.9%) worsened clinically and exhibited features of severe sepsis (n = 3), septic shock (n = 10), or systemic inflammatory response syndrome (SIRS) (n = 9).
3.4. Hematological Data
At diagnosis, the mean and median neutrophil counts were 0.13 and 0.06 × 10(9)/L (range: 0–0.48), respectively. In total, 59.2% of patients (n = 45) had neutrophil levels <0.1 × 10(9)/L. At the neutrophil decrease nadir, the mean and median neutrophil counts were 0.06 and 0 × 10(9)/L (range: 0–0.4). At nadir, a total of 93.4% of the patients (n = 71) displayed neutrophil levels <0.1 × 10(9)/L.
Overall, 28 patients (36.8%) exhibited isolated neutropenia, without any modification in red blood cells or platelets. The mean and median hemoglobin levels were 112.1 and 110 g/L (range: 74–153), respectively, while 50 patients (65.8%) suffered from anemia (hemoglobin <120 g/L). The mean and median platelet counts were 217 and 203 × 10(9)/L (range: 7–694), while 23 patients (11.3%) had thrombocytopenia (platelet count <100 × 10(9)/L).
Bone-marrow analysis (data available for n = 46) primarily detected myeloid hypocellularity with apparent cessation of myeloid precursor maturation (at the promyeloid stage) in 73.9% (n = 34).
3.5. Duration of Hematological Recovery and Response to Hematopoietic Growth Factors
The mean and median durations of hematological recovery (from initial neutropenia documentation to neutrophil count ≥1.5 × 10(9)/L) were 7.5 and 6 days (range: 2–21), respectively. The mean and median durations for achieving neutrophil counts ≥0.5 × 10(9)/L were 6.9 and 5 days (range: 1–21).
Granulocyte-colony stimulating factor (G-CSF) (administered subcutaneously at a fixed dose of 300 μg/day) was given to 45 patients (59.2%), particularly those with neutrophil counts <0.1 × 10(9)/L, severe clinical infectious features (e.g., collapse, septicemia, or extensive pneumonia), or renal failure. This hemopoetic growth factor (HGF) was administered for mean and median durations of 6.1 and 5 days (range: 1–18), respectively. For these 45 patients, the mean duration of hematological recovery was reduced to 0.7 days (range: 2–16), from 7.5 to 6.9 days (p = 0.089). The mean durations of antibiotherapy and hospitalization were not impacted when using HGF and consisted of 14.7 (range: 7–35) and 23.7 days (range: 5–74), respectively (all p > 0.4) (data not developed).
3.6. Management, Duration of Hospitalization and Outcome
All patients were hospitalized and treated immediately (within the first 24 h) with broad-spectrum parenteral antibiotherapy, mainly piperacilline (12 g/day) or cefotaxime (3 g/day), associated with netromycine (5 mg/Kg/day) or amikacine (15 mg/Kg/day) in sepsis cases, except in the event of β-lactam allergy or β-lactam-induced agranulocytosis. In a second step, the antimicrobial therapy was adapted according to microbes and antibiogram results, comprising mainly glycopeptide antibiotics and penems, associated with amphotericin or voriconazole in fungal infection cases. The mean and median durations of antibiotherapy were 18.3 and 15 days (range: 7–120), respectively. Forty-five patients (59.2%) were treated with HGF.
The mean and median hospitalization durations (available for 50 patients) were 26.3 and 12 days (range: 5–200), respectively. Eleven patients (21.6%; data available for 51 patients) required intensive care placement. In univariate analysis, there was no risk factor for admission to intensive care unit identified (p for all data not significant; data not developed).
Outcome was favorable in 89.5% of subjects (
n = 68), whereas eight patients (10.5%) died of either uncontrolled septic shock due to
Staphylococcus aureus and
Pseudomonas aeruginosa (
n = 5) or aggravation of deep abdominal and pelvic abscess (
n = 2) (
Table 5). The remaining patient died of hemorrhagic stroke (no relation with the febrile neutropenia). In univariate analysis, there was no risk factor for death identified (
p for all data not significant; data not developed). Four of these eight patients who died were treated with HGF.
4. Discussion
To our knowledge, this is one of the first studies focused on febrile neutropenia outside the oncology setting, with a consequent number of patients from a single center entered, with well-documented febrile neutropenia related to non-chemotherapy drugs (as “daily medication”) and managed using the same procedure. To our knowledge, two others studies have recently been published in such setting, one in non-immunocompromised children who attended an emergency department for febrile neutropenia [
11], and the second one in adult and child patients with febrile neutropenia [
12]. Our study’s novelty is that it is the first to investigate febrile neutropenia defined as established and documented idiosyncratic non-chemotherapy neutropenia.
Our patient diagnoses corresponded to both definitions of febrile neutropenia [
2] and idiosyncratic drug-induced agranulocytosis [
5,
6] (
Table 1). All patients exhibited unquestionable febrile neutropenia outside the oncology context: mean neutrophil counts of 0.06 × 10(9)/L (range: 0–0.4) at the neutrophil decrease nadir; 93.4% of patients with neutrophil levels <0.1 × 10(9)/L at nadir. In all cases, the principal diagnostic criterion for idiosyncratic agranulocytosis, namely complete hematological recovery following causative drug removal [
9], was fulfilled (except for the five patients who died of uncontrolled septic shock). As agranulocytosis is a life-threatening condition, no patient was re-challenged with the incriminated drug (“theoretical method of reference”).
In terms of the severity of infectious manifestations, clinical features observed in our population (
Table 3) did not differ from those observed in other idiosyncratic agranulocytosis series, involving hospitalized patients of all ages, [
5,
13]. In our population, the main clinical pictures upon hospitalization were pneumonia (18.4%), septicemia (14.5%), and septic shock (6.6%). Only one-third of patients exhibited isolated fever. In the course of hospitalization, 28.9% of our patients displayed features of severe sepsis, septic shock, and/or SIRS, whereas 21.6% required intensive care placement. As discussed elsewhere, the severity of the clinical manifestations we have reported from our series were probably accounted for by neutropenia severity (mean neutrophil counts of 0.06 × 10(9)/L) and perhaps by the patient types we had selected (patient referred to a referral center).
These reported clinical features differ from those observed in febrile neutropenia series involving chemotherapy-treated patients, particularly with myelosuppressive chemotherapy [
1,
2]. In this context, our patients did not exhibit deep invasive mycosis (except in one case of
Aspergillus sp. pneumonia and another case of maxillary sinus mucormycosis), as often reported in oncology and haematology patients [
14,
15]. Another observation is that our patients did not suffer from the following conditions: mucositis related to radio- or chemotherapy; deep immune deficiency related to malignant disorders associated to the neutropenia; exclusive “central” neutropenia (in opposition of “peripheral” neutropenia) related to partial bone marrow abolition (26.1% of our patients) [
1,
2]. It must also be mentioned that in most cases, a brutal installation of neutropenia was observed, which was of relatively short duration.
In our study, we observed a mortality rate of 10.5%. Of the eight death cases, only five elderly patients died of sepsis in relation to agranulocytosis, namely uncontrolled septic shock due to
Staphylococcus aureus and
Pseudomonas aeruginosa. The mortality in our population did not differ from the figures of other recent idiosyncratic drug-induced agranulocytosis series reporting rates from 5.4 to 11% [
5,
13]. This global mortality represents the upper bound of the mortality rates, despite our patients being relatively young (mean age: 52.2 years) and our team, as referral center, having wide experience of managing agranulocytosis. Explanations for these discrepancies may be the severity of the infectious pictures we encountered, in addition to that 86% of patients presented underlying diseases that were often not stabilized, especially cardiac, neuro-psychiatric, and renal disorders. In the Julia et al. study, renal failure was considered a poor prognostic indicator reflecting severe infections, in association with a neutrophil count <0.1 × 10(9)/L [
13]. In addition, Maloisel et al. have demonstrated that in the event of severe neutropenia (absolute neutrophil count <0.1 × 10(9)/L), the prognosis of idiosyncratic agranulocytosis to be impacted by the following factors: age (75 years-old), infection severity (septicemia and septic shock), and comorbidities (particularly renal failure) [
16].
This mortality rate slightly differs from figures published on chemotherapy-related febrile neutropenia series [
1,
2]. In oncology or hematology settings, febrile neutropenia has been shown responsible for considerable morbidity, given that 20–30% of patients exhibited complications requiring in-hospital management, with an overall in-hospital mortality of about 10% [
1,
2]. In this context, the mortality figures reported appear directly impacted by the prognosis of the underlying cancer. There is a clear relationship between the severity of neutropenia (neutrophil count <0.5 × 10(9)/L) and intensity of chemotherapy. In terms of risk for febrile neutropenia, the different therapeutic regimens have been classified classified as high-risk (>20%), intermediate-risk (10–20%), or low-risk (<10%) [
1,
2]. Thus, mortality was shown to vary according to the Multinational Association of Supportive Care in Cancer (MASCC) prognostic index: lower than 5% if the MASCC score was ≥21, but possibly as high as 40% if the MASCC score was <15 [
2]. Other factors with a similar role are: old age, several comorbidities, and performance status (OMS or Charlson score) [
1,
17]. To our knowledge, these factors (dependent of host) have not been well-studied in the context of idiosyncratic agranulocytosis that is unrelated to chemotherapy, particularly febrile neutropenia cases.
The above-mentioned considerations should be instrumental in deciding whether a chemotherapy-treated patient should receive primary or secondary HGF prophylaxis to decrease the potential febrile neutropenia risk [
1,
4]. In our study, we were unable to identify a risk factor for death. HGF use (in curative perspective) did not appear to alter either patient disease progression or mortality. So far, no data are available for HGF prophylaxis, except for the few published case-reports of clozapine-induced agranulocytosis [
18].
Outcome was favorable in 89.5% of cases. In our center, all patients underwent established care procedures (for details, see [
6,
8]) (
Table 6). This protocol has been modeled based on that pertaining to febrile neutropenia management in the oncology setting [
1]. In our opinion, this may account for our good results despite the severity of patient clinical manifestations. In our experience, appropriate management of septic complications of idiosyncratic agranulocytosis, using both broad-spectrum antibiotherapy and HGF, may improve the condition’s prognosis [
6,
16].
A faster hematological non-significant recovery (neutrophil count >1.5 × 10
9/L) was observed in the HGF group: −0.7 days (from 7.5 to 6.9 days) (
p = 0.089). Nevertheless, there were no other improvements observed in these patients in relation with HGF therapy, regarding antibiotherapy and hospitalization duration. This was also the case in the oncology setting. No impact was documented in the event of HGF administration for neutropenic febrile patient (“curative” use of HGF) [
4]. While these results do not seem to be in line with those previously reported by our research team [
6], they are consistent with those originating from larger HGF studies involving adult idiosyncratic drug-induced agranulocytosis patients [
19,
20]. For a number of hematologists, HGF usefulness in this setting is still a matter of debate. In line with this debate, the only available prospective randomized study (based on 24 patients with antithyroid-related IDIA) did not confirm the benefits of administering G-CSF [
21]. It should, however, be mentioned that this negative study result may be accounted for by inappropriate G-CSF doses (100–200 μg/day) [
6].
Of note is also that all our patients with febrile neutropenia were promptly hospitalized. For us, it appears mandatory that this consensual recommendation be an integral part of idiosyncratic agranulocytosis management [
5,
6]. This should, however, not be the case for chemotherapy-induced febrile neutropenia. In the oncology setting, several of these patients may be treated on an ambulant basis (outside the hospital) [
1,
2], whereas all patients classified as “high-risk” by MASCC (MASCC < 15) or meeting clinical severity criteria should initially be admitted to the hospital for empirical antibiotic therapy if they are not already hospitalized. Carefully selected “low-risk” patients (e.g., MASCC score > 21) may be candidates for empirical antibiotic therapy on an outpatient basis [
2]. To our knowledge, such appropriately-structured and score- or index-guided management is not yet available for idiosyncratic febrile neutropenia cases, although each clinician may have personal experience with such management in selected young patients without comorbidity (e.g., anti-thyroid drug-induced agranulocytosis). Concerning home management, only few retrospective and poorly-designed reports have been published in the scientific literature [
5,
6].
Our work further highlights the “classical” causative drugs: antibiotics (37.4%); antithyroid drugs (17.2%); neuroleptic and anti-epileptic agents (13.1%); non-steroidal anti-inflammatory agents and analgesics (8%); and platelet aggregation inhibitors (8%), especially ticlopidine. Everyday medicines may thus be implicated in severe accidents. It is, thus, mandatory to appropriately inform and train all practitioners, including those practicing in the city. Our present data were in accordance with those published by Shapiro et al. [
22] and van der Klauw et al. [
23,
24]. In our study, antibiotics were at the top of offending drug classes. It should, however, be noted that in clinical practice, it may often prove challenging to formally attribute neutropenia to a specific drug rather than the infection for which the drug was initiated [
5,
25]. In our study, the reported cases all fully met the required criteria, as defined in the scientific literature.
Our study does not enable any conclusion to be drawn about the usefulness of blood cell count monitoring for certain medicines at risk (e.g., antithyroid drugs, ticlopidine, and clozapine), as proposed elsewhere [
5,
6]. Although patients experiencing idiosyncratic agranulocytosis may initially be asymptomatic or suffer from isolated fever, which was the case in our patients, neutropenia often reflects the onset of severe sepsis depending on its severity and depth [
5,
6]. Thus, the usefulness of surveillance may be questioned, especially for drugs with a low incidence of idiosyncratic neutropenia.
Our study exhibited several limitations. Chiefly, the data originated from a population covering a period of over 30 years. There may thus be heterogeneity as to the causative drug and possibly the management process. Moreover, all patients were referrals. On the other hand, our study exhibited several strengths, in that it was the first to investigate febrile neutropenia in the setting of established and documented idiosyncratic non-chemotherapy-induced neutropenia. In addition, this study was conducted in a single center, with experienced physicians well- accustomed to managing neutropenia and agranulocytosis. Therefore, it appears difficult to generalize any results and conclusions, and caution is required when extrapolating the findings outside this single center. The small sample size, with only 76 patients included, limited the study’s power to detect any meaningful statistical and clinical patterns. The statistical analyses were thus mostly exploratory without adjusting for any potential confounding factors, and the results could be potentially biased by confounding factors. A larger-scale study with data originating from multiple centers is necessary to confirm our findings, in addition to a sophisticated statistical analysis plan.