Deciphering cytopenias in internal medicine: a single-center observational study

Cytopenia is a common finding in patients admitted to internal medicine wards and the clinical workup may be long and time-consuming. In this single-center observational study, we analyzed a series of 151 inpatients who received hematologist referral due to cytopenia observed during hospital admission. Patients were mainly elderly (median 71 years, 15–96) and 87% had at least one comorbidity. Anemia was the most common cytopenia (91%), followed by thrombocytopenia (51%), and neutropenia (22%); 73 (48%) patients had a bicytopenia and 5 (3%) pancytopenia. Cytopenias were mainly severe, 66% of cases required RBC transfusions, and 21% platelet pools. During a median hospital stay of 15 days (1–166), 53 subjects (35%) received a hematologic discharge diagnosis, whilst the two-thirds had secondary cytopenia mainly due to associated comorbidities. Only about 34% of 2,728 diagnostic tests performed (including laboratory, imaging, and histology) clearly informed the discharge diagnosis in this heterogenous setting. Specifically, bone-marrow evaluation indicated in 46 (30%) patients, was diagnostic in 32 (69.6%). Eleven percent of patients died due to progression of the oncohematologic disease (29%), sepsis (24%), and solid tumor progression (24%). In conclusion, cytopenias in the internal medicine setting are mainly severe, more frequently secondary to associated comorbidities (2/3 of patients) and deserve proper workup before second/third-level tests (immune-hematological assays and CT scan or PET and bone-marrow evaluation, respectively). Supplementary Information The online version contains supplementary material available at 10.1007/s11739-023-03517-z.


Introduction
Cytopenia represents a common clinical finding in patients admitted to internal medicine and surgical wards [1 -3].The underlying causes are often primarily non-hematological and include nutrients deficiencies, chronic diseases, inflammatory state, and cancer [4,5].The differential diagnosis may be challenging, since most patients are elderly and carry several comorbidities that may further confound the picture [6].Hematologist consultation may lead to a number of specialistic laboratory and instrumental investigations.The utility and reliability of the latter during admission may be questionable, particularly due to concomitant acute conditions, such infections, and drug use.In addition, the lack of predictors of hematologic diagnoses in these patients does not aid to discriminate those needing second-level tests, such as bone-marrow (BM) aspirate and trephine.Whilst computer tomography (CT) and positron emission tomography (PET) scans are largely and autonomously used by internal medicine specialists, BM evaluation is generally indicated by the hematologist and its interpretation carries several drawbacks [7].In this study, we analyzed a series of inpatients who received hematologist referral due to cytopenia observed during hospital admission.We focused on the laboratory and radiological investigations performed since admission until diagnosis, including those asked by the hematologist, to identify possible indicators of underlying hematologic condition.Furthermore, we analyzed mortality rate and causes among cytopenic patients and evaluated possible clinical and laboratory associations.

Design of the study
This is a retrospective cohort study conducted between January 2018 and December 2019 at IRCCS Fondazione Ca' Granda-Ospedale Maggiore Policlinico (Milan, Italy).The study was approved by the local Ethical Committee as a substudy of the CYTOPAN observational trial [NCT05931718] and was conducted according to the Helsinki Declaration.
Patients admitted to non-hematological wards who were referred to the hematologist due to anemia (Hb < 12 g/dl), thrombocytopenia (platelets, PLT < 100.000/mmc) or neutropenia (absolute neutrophil count, ANC < 1500/mmc) without a known hematological diagnosis were included.Patients were stratified according to age (< 65 years, between 65 and 79, and ≥ 80) and according to the severity of the cytopenia [8,9].

Clinical and hematologic workup
Hematologic parameters, including hemolytic markers (LDH, haptoglobin, total and fractionated bilirubin, and absolute reticulocytes), immune-hematologic evaluation (anti-PLT autoantibodies, direct and indirect antiglobulin test, DAT and IAT, ANA, ENA, anti-DNA, and antiphospholipid antibodies), serology for major hepatotropic viruses (HBV, HCV), HIV, Herpes viruses (CMV, EBV) and Parvovirus B19, as well as the isolation of bacterial or fungi on cultural exams, were recorded.Furthermore, BM evaluation (aspirate and trephine) and imaging studies (ultrasound sonography, US, CT, PET) reports performed from initial admission to cytopenia diagnosis were registered.Presence of comorbidities was also collected.The following comorbidities were considered: cardiovascular disease (congestive heart failure, arrhythmias, ischemic cardiomyopathy, other cardiomyopathies, arterial hypertension, peripheral vasculopathy, valvulopathy, presence of intravascular devices), kidney disease, hepatic disease (acute hepatitis, chronic hepatopathy, cirrhosis, hepatocellular carcinoma), autoimmune disease, diabetes mellitus, thyroid disease, sepsis, and thrombosis.

Treatment and outcome
All therapies administered during admission were recorded with particular attention to those indicated for the cytopenia [10][11][12].The final diagnosis was registered as reported by the discharging physician and classified as hematologic or non-hematologic, and the former distinguished in oncologic and non-oncologic.Hematologic diagnoses were then classified according to current WHO classification of myeloid and lymphoid diseases [13].A comparison of baseline demographic, clinical and laboratory features, including cytopenia type and severity and presence of comorbidities, was performed between patients receiving or not a hematological diagnosis.
Occurrence of death and its cause were also collected.Mortality analysis included the evaluation of clinical and laboratory variables (age and gender, diagnosis, hematologic parameters, comorbidities, and therapies) possibly associated with fatal outcome.

Statistical analysis
Descriptive statistics was applied to describe the distribution of variables among the different subgroups of patients.Student t test was used to compare continuous variables, Chi-square and Fisher exact test were applied to evaluate categorical ones.The sensitivity, specificity and positive predictive value for DAT and anti-PLT antibodies was also calculated.

Demographics
A total of 170 hematologic consultations were evaluated: 151 were included since had been performed for new onset cytopenia, whilst 19 were excluded since regarded patients already on hematologic follow-up (Supplementary Fig. 1).Patients were mainly elderly (median age 71 years, range 15-96; 33% of patients > 80 and 32% between 65 and 80 years), and male to female ratio was 1.25 (Table 1).Regarding medical history, 87% of patients had at least one comorbidity, mainly cardiologic (59%) and liver diseases (39%), followed by kidney disease, diabetes, and autoimmune diseases (about 20% each) (Supplementary Table 1).

Diagnostic tests
A total of 2,728 diagnostic tests were performed, resulting in about 18 tests per patient, including laboratory and imaging investigations, and bone-marrow evaluations.Overall, only about 34% of tests reported a result clearly informing the discharge diagnosis.Table 3 shows the most relevant tests.Regarding immune-hematological investigations, 6 patients (12%) among those tested showed positive DAT and 6 (12%) positive IAT.All DAT positive patients also presented altered hemolytic markers and received a discharge diagnosis of AIHA.This resulted in a sensitivity of 100% and a specificity of 97%.Anti-PLTs antibodies were detectable in 27 patients (56% of tested cases), of whom only 7 received a final diagnosis of ITP.The calculated sensitivity and specificity were 86% and 49%, respectively.Anti-neutrophil antibodies were positive in only 1 patient (25% of tested cases).In addition, 17 patients (34% of those tested) showed ANA positivity, 4 (10%) positive ENA antibodies, and 2 (8%) anti-phospholipid antibodies.Concerning imaging, splenomegaly was found at US in 46 (31%) patients, 17 (32%) of whom received a oncohematologic diagnosis at discharge.Contrast enhanced whole body CT scan was performed in 71 (37%) patients and infections were the most frequent finding (28%), followed by splenomegaly (23%).Only 20 (13%) patients underwent PET scan: lymph node hypercaptation was the most common result, which was equally distributed between hematologic and non-hematologic patients.Spleen and bone-marrow hypercaptation were also frequent (35% and 30%, respectively).The finding of lymphoadenopathy and splenomegaly at CT scan was significantly more frequent in patients receiving a hematologic diagnosis (50% vs 17.6%, p = 0.04).Similarly, there was a trend for higher spleen and bone hypercaptation at PET examination in this group.
Figure 1 displays the different tests performed according to the frequency of alteration and the level of investigation: first level (blood counts, nutrients, serology for common viral infections, and abdomen ultrasound), second level (immune-hematology tests, organ and non-organ specific autoantibodies, CT and PET scan), and third level (bonemarrow evaluation).
Causes of death included progression of the oncohematologic disease (29%), sepsis, and solid tumor progression (4 patients, 24% each).Only one patient died for severe hemolytic disease.Mortality was significantly associated with sepsis (41% septic vs 11% not, p = 0.005) and need of platelets transfusions (59% transfusion dependent vs 15% not, p < 0.001).Further univariate and multivariate analyses did not highlight significant associations.Subjects' engagement for outpatient follow-up was significantly more frequent in hematologic patients than in nonhematologic ones (p ≤ 0.0001).

Discussion
Here we describe a large series of inpatients evaluated due to peripheral cytopenias admitted to non-hematological wards and show that they are mainly elderly (twothirds > 60 years) and carry a burden of comorbidities (87% of cases) mainly involving chronic kidney and liver disease.Cytopenias were often severe, one-third of patients required transfusions, and 11% died.After a median hospital stay of 15 days, only one-third of subjects received a discharge hematological diagnosis, equally distributed between onco-and non-oncological ones.Hematological patients showed more frequently neutropenia and thrombocytopenia as compared to nonhematological ones, whilst anemia and hemolytic markers alteration did not display significant association.
Although mortality was not significantly associated with a hematological diagnosis, up to one-third of subjects died during progression of oncohematologic disease and one due to fatal autoimmune hemolytic anemia.This data appears remarkable considering the short-term observation, highlighting the likely impact of cytopenias on survival [14,15].
Figure 2 depicts a proposed work up for cytopenias in the internal medicine setting, including first-level tests (CBC, nutrients, and organ functions), second-level (immunehematologic and imaging), and third-level ones (BM evaluation).In our series, only one-third of thousands of tests performed were informative on the final discharge diagnosis.If "first-level" tests, mainly aimed at excluding secondary cytopenias, are generally not expensive, widely available, and autonomously ordered by the internal medicine specialist, others may be considered at a second time.These include immunohematologic tests, whose positivity was high, ranging from 4% for anti-phospholipid antibodies to 56% for anti-PLT antibodies, as in other series [16,17].Only DAT appeared highly sensitive and specific for AIHA diagnosis and is, therefore, recommended in case of anemia with altered hemolytic markers [18][19][20].Contrarily, anti-PLT autoantibodies test confirmed to be less specific and is not routinely advised [21,22].Second-level imaging (CT and PET scans) appears valuable in the evaluation of cytopenic patients [23], although the main finding was the presence of infectious foci, and splenomegaly and lymph node enlargement were equally distributed among oncologic and benign diagnoses.Hematologists indicated BM aspirate and trephine in only 1/3 of cases, and this led to hematological diagnosis in about 60% of patients with persistent cytopenia of unknown cause.Notably, the most frequent finding in this setting was myelodysplasia that may be difficult to quantify, may be partly age-related [24][25][26][27][28], or associated with nutrients deficiencies that should be corrected before performing the test.Since most patients were elderly and with comorbidities we could not find specific factors associated with uninformative BM evaluation.In general, BM evaluation should be pursued in those patients, where it might significantly impact the clinical management and after all first-and second-level tests have been already performed.Study limitations include the retrospective nature of the study and the limited number of patients, although the detailed clinical data collected allowed to depict a real-world picture of a cytopenic internal medicine inpatient population.
In conclusion, we confirm that secondary cytopenias are more frequent in the internal medicine setting (2/3 of patients) and deserve proper workup before second/ third-level hematologic tests [29].The burden of poorly informative investigations remains an open issue depending on cytopenia severity, clinician' expertise, as well as on availability and costs of the various tests.

Fig. 1
Fig. 1 Different diagnostic tests for cytopenias according to percentage of alteration and level of investigation.Number and percentage of alteration or positivity of the test are reported in brackets

Table 1
Demographics and reason for admission *p < 0.05 Hb hemoglobin, PLTs platelets, ANC absolute neutrophil count; values are expressed as N(%) unless otherwise specified

Table 2
Discharge diagnosis and outcome