Quantitative Evaluation of Noncontrast Computed ...

1 downloads 0 Views 593KB Size Report
attenuation of dural venous sinuses correlates with hemoglobin/hematocrit and to determine if the degree of anemia can be predicted by quantitative analysis of ...
ORIGINAL ARTICLE

Quantitative Evaluation of Noncontrast Computed Tomography of the Head for Assessment of Anemia Ammar A. Chaudhry, MD,* Maryam Gul, MD,† Abbas Chaudhry, PharmD,* Mubashir Sheikh, MS,* and Jared Dunkin, MD* Background and Purpose: Anemia is an underdiagnosed clinical entity with significant mortality and morbidity. We aimed to assess whether attenuation of dural venous sinuses correlates with hemoglobin/hematocrit and to determine if the degree of anemia can be predicted by quantitative analysis of unenhanced computed tomography (CT) of the head. Materials and Methods: This is an institutional review board– approved retrospective study including 500 patients who underwent emergency department investigation for potential central nervous system etiology of their symptoms with unenhanced CT head at a tertiary care center. Computed tomographic attenuation values were obtained by 2 independent readers, whereas 2 separate investigators collected clinical data. Regression analyses were performed to evaluate the strength of correlation and the predictability of anemia and its severity on unenhanced CT. Receiver operating characteristic curve analyses were performed to evaluate sensitivity, specificity, as well as positive and negative predictive values. Results: A total of 243 met the inclusion criteria, and attenuation values for all the dural venous sinuses were averaged and categorized according to hemoglobin values of less than 8, 8 to 10, 10 to 14, and greater than 14. Mean CT attenuation values for both readers were 36.30, 42.35, 47.99, and 53.25 Hounsfield units. Regression analysis revealed the highest positive correlation of hemoglobin/hematocrit with attenuation at the confluence of sinuses with R2 value of 0.63 and 0.60. The sensitivity, specificity, and negative predictive value of detecting hemoglobin of less than 10 at confluence of sinuses were 91.2%, 88.5%, and 98.6%, respectively. Interobserver agreement was found to be good (0.64) using the κ statistic. Conclusions: Our study substantiates direct positive correlation between CT attenuation of dural venous sinuses and hemoglobin/hematocrit, with strongest correlation at the confluence of sinuses with good sensitivity, specificity, and negative predictive value. Key Words: anemia, computed tomography, dural venous sinus (J Comput Assist Tomogr 2015;00: 00–00)

T

he World Health Organization defines anemia as hemoglobin (Hb) less than 13 g/dL in males and less than 12 g/dL in females.1 Anemia is a widespread clinical entity with significant mortality and morbidity, which remains underdiagnosed and untreated.1,2 The prevalence of anemia increases with age, with global incidence of 24.8% (1.62 billion).3 In the United States, From the *Department of Radiology, Stony Brook University Medical Center, Stony Brook; and †Department of Internal Medicine, Winthrop University Hospital, Mineola, NY. Received for publication April 30, 2015; accepted June 2, 2015. Correspondence to: Ammar A. Chaudhry, MD, Division of Neuroradiology, Department of Radiology, Johns Hopkins Medicine, 600 North Wolfe Street, Phipps-B100, Baltimore, MD 21287 (e‐mail: [email protected]). A.C. is a medical student from the Department of Radiology, Stony Brook University Medical Center, Stony Brook, NY. The authors declare no conflict of interest. The study was part of an oral presentation at ASNR 2014 annual meeting in Montreal, Quebec, Canada. Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved. DOI: 10.1097/RCT.0000000000000306

J Comput Assist Tomogr • Volume 00, Number 00, Month 2015

the prevalence of anemia ranges from 9.2% to 23.9% in males and 8.1% to 24.7% in females.1 However, it is much higher in elderly nursing home patients (48%–63%) and hospitalized patients (37%).1,3–5 Anemic patients are more likely to be hospitalized, have increased lengths of hospital stays, have increased disability resulting in greater health care costs, and have increased overall mortality.1,2,5 Given this, early detection and prompt treatment are imperative to avoid poor clinical outcomes.1,5,6 Anemia can present with vague neurologic symptoms such as altered mental status, headache, dizziness, and syncope, which are associated with increased incidence of falls.6–8 Neurologic symptoms are often the impetus for ordering a large number of head computed tomographies (CTs) in the emergency department setting. The absence of intracranial pathology, while comforting, can also be challenging for patients and physicians who are working to identify an underlying cause for the patient’s symptoms. Therefore, having the ability to suggest anemia from an otherwise unremarkable head CT result can facilitate patient care as the presenting symptoms may frequently be the result of anemia. In addition, if we can grade the severity of anemia on CT and identify patients with critical anemia (Hb < 8 g/dL),9 we can immediately communicate the findings with clinicians, thus expediting patient care. The linear relationship between Hb levels and CT attenuation (as measured in Hounsfield units [HU]) has previously been established and so has the diagnosis of anemia to be reliably made on CT of the thorax and abdomen.6,10–12 In our practice, we have observed a similar trend with regard to CT attenuation of dural venous sinuses (DVS) with patients’ Hb and hematocrit (Hct) on a noncontrast CT. Previous studies have established this direct positive correlation between attenuation of DVS on unenhanced CT with Hb and Hct.13–21 However, for various reasons (small sample size, scanner/reconstruction protocol variability, etc), none of the previous studies have been able to establish statistically significant data regarding sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV) of detecting anemia on unenhanced CT head. In addition, these studies failed to demonstrate predictability of severity of anemia on noncontrast CT. We aimed to evaluate the strength of correlation between attenuation of intracranial DVS and patients’ Hb and Hct values. If there is a correlation, we aimed to assess the sensitivity, specificity, PPV, and NPV of CT attenuation values of DVS in diagnosing anemia. Furthermore, we aimed to assess whether CT attenuation of DVS attenuation values can suggest the severity of anemia because such information could add critical information to patient care.

MATERIALS AND METHODS Study Group In this institutional review board–approved retrospective study, 500 consecutive patients (290 males, 210 females) who underwent emergency department investigation for potential central nervous system etiology of their symptoms (presyncope, www.jcat.org

Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. This paper can be cited using the date of access and the unique DOI number which can be found in the footnotes.

1

Chaudhry et al

J Comput Assist Tomogr • Volume 00, Number 00, Month 2015

FIGURE 1. Axial (A), sagittal (B), and coronal (C) images on this noncontrast CT of the head reveal global hypoattenuation of the DVS with CT attenuation less than 36, suggestive of anemia. This was confirmed with complete blood count revealing Hb of 4.2 g/dL and Hct of 11.6, diagnostic of severe anemia.

syncope, headache, altered mental status, blurry vision, etc) with unenhanced CT head at a tertiary care center from November 2012 through February 2013 were randomly selected. Patients were included irrespective of age, sex, or clinical indication. Patients were excluded from the study because of limitations including motion or streak artifact, if patient received iodinated contrast in 48 hours before the head CT or if there was history or imaging evidence of trauma (intracranial hemorrhage, fractures, foreign bodies, etc), postsurgical changes (ventricular shunts, metallic devices, etc), intracranial neoplasms, evidence of cerebrovascular accident, or marked calcifications along the DVS. Studies were also excluded because of the presence of additional diagnosis that may at least partially explain patients’ neurologic symptoms (metabolic disturbances [acidemia, hyponatremia/ hypernatremia, ketoacidosis, etc], infection/sepsis, malignancy, etc) and unavailability of Hb and Hct values obtained within 24 hours of the CT scan.

Imaging Protocol All noncontrast CT studies of the head were volume scans performed on a 320-detector row system (Aquilion One; Toshiba Medical Systems, Nasu, Japan), with detector width of 16 cm per gantry rotation, total cone angle of 15.2 degrees, and minimum gantry rotation time of 350 milliseconds. Image reconstruction is performed by using a filtered back-projection–type cone beam reconstruction algorithm (ConeXact; Toshiba Medical Systems). Scan parameters for this protocol include peak tube voltage 120 kV, tube current of 320 mA, rotation time of 1.0 second, and axial slice thickness of 0.5 mm with reconstruction interval of 0.25 mm. A primary DICOM [Digital Imaging and Communications in Medicine] volume of 640 axial slices of 0.5-mm section thickness and 0.25-mm slice increment resulted. Radiation exposure as stated by the software of the console on average for a scan length of 16 cm was approximately 39.9 mGy (dose-length product, 558.7 mGy-cm). From the primary data, standard axial, sagittal, and coronal images were reconstructed each with a slice thickness of 2.5 mm. Studies without all 3 reconstructions were excluded from the study.

resident. Computed tomographic attenuation values were measured from the superior sagittal sinus (SSS), confluence of sinuses (CoS), right transverse sinus (RTS), and left transverse sinus (LTS) using preset soft tissue window settings (window width, 80 HU; window level, 40 HU). Both readers were aware of the study objectives, both reviewed the studies in different locations at different times, and both were blinded to patients’ clinical data. For each case, regions of interest of approximately 3 mm in diameter with SD of less than 5 units were placed on the SSS, CoS, and bilateral transverse sinuses at the widest cross-sectional interval in each orthogonal plane (coronal, sagittal, and axial), ensuring sinus wall was not included in the regions of interest (Fig. 1). In each case, the 3 values obtained from each orthogonal place for each DVS were averaged and documented. A third and fourth independent investigator (both medical students) evaluated patients’ medical records and collected clinical parameters including clinical history, indication for CT, laboratory values including Hb (range, normal 14–18 g/dL) and Hct (range, 42.0–52.0). Clinical definition of anemia based on Hb concentration and perceived severity of anemia varies in the literature; this is not surprising given that, like many conditions, a diagnosis of anemia depends on the patient population and the different test tools used to measure Hb values.1,3,4,22 Transfusion parameters also

Study Design Quantitative image analysis was performed by 2 independent readers. The first reader, reader A, is a fellowship-trained attending neuroradiologist with more than 6 years of experience as an attending. The second reader, reader B, is a third-year radiology

2

www.jcat.org

FIGURE 2. Histogram shows normal distribution of Hb values in the study cohort. © 2015 Wolters Kluwer Health, Inc. All rights reserved.

Copyright © 2015 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. This paper can be cited using the date of access and the unique DOI number which can be found in the footnotes.

J Comput Assist Tomogr • Volume 00, Number 00, Month 2015

Non-Contrast CT Head for Assessment of Anemia

TABLE 1. Mean CT Attenuation Values of DVS With Patients Stratified According to Severity of Anemia Mean Age 58.98924731 Hb Range No. Patients CT Attenuation, Mean (SD) SSS, HU RTS, HU LTS, HU CoS, HU Mean (SD) HU CT Attenuation, Median Hb Range No. Patients SSS, HU RTS, HU LTS, HU CoS, HU Median HU

Age Range 5–99 Hb < 8 g/dL 10

Female 133 Hb, 8–10 g/dL 64

34.3 (6.7) 40.0 (11.9) 37.1 (8.4) 33.8 (5.6) 36.3 (8.2)

42.4 (3.7) 43.8 (4.6) 42.9 (4.1) 40.2 (4.2) 42.4 (4.1)

Hb < 8 g/dL 10 35.00 37.50 40.50 34.00 35.7

Hb, 8–10 g/dL 64 42.50 43.50 43.00 40.50 42.4

vary within medical specialties and clinical setting.9,22–28 The most widely accepted anemia grading and transfusion parameter are from the American Association of Blood Bank (AABB).9 According to AABB, Hb less than 8 is considered severe almost always requiring blood transfusion, Hb of 8 to 10 is moderate, which may or may not require blood transfusion, depending on comorbid conditions, and Hb of 10 to 14 represents mild anemia, which generally does not require blood transfusion.3,9 Patients were stratified based on the aforementioned AABB anemia severity scale.

Statistical Analysis Statistical analyses were performed using MedCalc for Windows, version 12.5 (MedCalc Software, Ostend). After controlling for patient sex, measures of central tendencies were performed, and mean CT attenuation values (HU) for each DVS within the aforementioned subgroups was calculated along with SD. In addition, Pearson correlation coefficient (PCC) was used to evaluate the strength of correlation of CT attenuation values of the DVS with Hb and Hct. By using the aforementioned anemia severity scale, receiver operating characteristic (ROC) curve analysis was performed, and sensitivity, specificity, PPV, and NPV for the diagnosis of anemia from DVS CT attenuation were obtained. Linear and multiple regression analyses including analysis of variance were performed to assess whether severity of anemia can be predicted based on CT attenuation values. To account for reader variability, κ statistic was used to assess for interobserver agreement.

RESULTS Of the 500 patients, 243 (110 males, 133 females; mean age, 59 years; range, 5–99 years) met the inclusion criteria. A histogram of the patients’ Hb and Hct shows a normal distribution pattern for the subjects included in the study (Fig. 2). Pearson correlation coefficient (r) between patients’ Hb and Hct was 0.98 (P < 0.0001; 95 confidence interval [CI], 0.969–0.981), confirming strong positive linear correlation. A strong positive direct correlation between CT attenuation of CoS and SSS was noted (r = 0.90), (P < 0.0001; 95 CI, 0.866–0.917), suggesting © 2015 Wolters Kluwer Health, Inc. All rights reserved.

Male 110 Hb, 10–14 g/dL 250

Hb > 14 g/dL 162

47.4 (4.5) 50.6 (4.6) 49.1 (5.2) 44.8 (4.1) 47.9 (4.6) Hb, 10–14 g/dL 250 47.50 50.50 49.25 45.25 48.1

52.5 (4.35) 55.5 (5.3) 54.9 (6.3) 50.1 (4.1) 53.2 (5.0) Hb > 14 g/dL 162 52.00 55.50 55.00 50.00 53.1

TABLE 2. Pearson Correlation Coefficient (r value) Between DVS With Hb and Hct for Both Readers Combined, Readers A and B PCC (r) Between CT Attenuation of DVS With Hb and Hct Combined readers (Hb) CoS SSS LTS RTS Combined readers (Hct) CoS SSS LTS RTS Reader A (Hb) CoS SSS LTS RTS Reader A (Hct) CoS SSS LTS RTS Reader B (Hb) CoS SSS LTS RTS Reader B (Hct) CoS SSS LTS RTS

r 0.7053 0.6764 0.6465 0.6614

95% CI 0.6577–0.7474 0.6250–0.7219 0.5916–0.6955 0.6082–0.7087

P