**1. Introduction**

Orbital space-occupying lesions comprise a wide range of benign and malignant masses [1]. The top eight ocular adnexal lesions include lymphoid tumors, inflammatory disease, cavernous hemangioma, lymphangioma, meningioma, optic nerve glioma, metastatic breast cancer and capillary

hemangioma [2]. Several studies indicate that ocular adnexal lymphomas almost account for up to half of all malignant orbital lesions in adults [2–4].

Immunoglobulin G4–related disease—which can involve any organs including orbital structures is a systemic fibroinflammatory condition due to tissue infiltration by immunoglobulin G4 (IgG4) plasma cells [5]. IgG4-related disease in orbit tends to form tumor-like lesions, which are difficult to differentiate from intraorbital lymphoma because both are tumors rich in lymphoplasmacytic infiltration [6]. Lymphoma, one of the most common orbital malignancies in adults [3,4], needs radiotherapy, systemic chemotherapy and/or surgery [7,8]. In contrast, approximately 90% of patients with IgG4-related orbital disease (IgG4-ROD) respond well to steroid therapy [5] Therefore, rapid and accurate diagnosis of IgG4-ROD to help patients receive early steroid treatment is critical.

IgG4-related disease diagnostic criteria commonly follow: a serum IgG4 concentration higher than 135 mg/dL (milligrams per deciliter) [9], the ratio of IgG4-positive/immunoglobulin G (IgG)-positive plasma cell (IgG4+/IgG+ ratio) is >40% or IgG4+ cells > 10/high-powered field of biopsy sample [9]. However, up to 40% of patients with IgG4-related disease may have serum IgG4 levels within the normal range [10]. Pathology and immunohistochemistry remain the gold standard for accurate diagnosis of IgG4-related disease [5,11].

Furthermore, orbital neoplasm rupture via biopsy may lead to tumor seeding and poor prognosis [12]. Integrating clinical findings, serologic data and radiological features is important to establish the prebiopsy diagnosis of IgG4-ROD [12]. Computed tomography (CT) scans provide rapid high-resolution images of orbits for radiological feature extraction [13]. As such, this study integrates CT qualitative and quantitative (Hounsfield unit density) features, serum IgG, and IgG4 levels to differentiate IgG4-ROD from orbital lymphoma (OL).

#### **2. Material and Methods**

#### *2.1. Patients*

The Institutional Review Board of Taipei Veterans General Hospital (TVGH) approved this study to waive informed consent because of the retrospective nature of the research.

A doctor (SYY) blinded from the research hypothesis searched pathologic results from the pathology database at TVGH using the keywords "orbit" or "orbital" for cases from 1 January 2010, to 31 May 2018.

The research returned 178 patients with orbital lesions and pathologic results. Thirteen (7%) of 178 patients had multiple orbital lesions (≥2): one (7.7%) of the thirteen patients with lung carcinoid tumors and multiple metastases in the right orbital cavity and 12 (92.3%) patients with lymphoplasmacyte-rich lesions (5 patients with orbital lymphoma; 7 patients with IgG4-ROD). The other 165 (93%) of the 178 patients showed only one lesion in the orbital cavity or eyelid. Sixteen (9%) of the 178 patients had orbital lymphoma (OL) and 9 (5%) patients had IgG4-related orbital disease (IgG4-ROD).

Among these patients, this study only considered patients who had OL or IgG4-ROD with prebiopsy precontrast and postcontrast orbital CT scans, serum IgG4 levels and serum IgG levels and excluded those were younger than 20 years of age or pregnant cases or those lacked prebiopsy CT and serologic data.

As a result, 25 patients pathologically diagnosed as OL (16 patients) or IgG4-ROD (9 patients) met the inclusion criteria and had prebiopsy orbital CT scans, serum IgG4 levels and serum IgG levels. All patients met the eligibility criteria. We enrolled 25 patients to collect and analyze demographic data, symptoms and signs, past medical histories, CT qualitative and quantitative (Hounsfield unit density) features of orbital lesions, serum IgG levels and serum IgG4 levels of patients with IgG4-ROD from those of patients with OL.

#### *2.2. CT Imaging Techniques*

This study examined orbital CT images taken by a multiple-detector computed tomography (MDCT) scanner for the selected 25 patients. MDCT scanners of orbit or face included iCT 256 (256-slice, *n* = 5), Philips Healthcare, Cleveland, OH, USA, Somatom Sensation 16 (16-slice, *n* = 4), Siemens Healthcare, Forchheim, Germany, ECLOS Hitachi Medical Corporation (16-slice, *n* = 1), Tokyo, Japan, and Aquilion 64, Toshiba Medical Systems (64-slice, *n* = 15), Tochigi, Japan. Orbital CT scans were obtained with or without an intravenous contrast medium, which included iobitridol (Xenetix 350; Guerbet, Rue Jean Chaptal, Aulnary-sous-Bios, France, 350 mg I [Iodine]/mL [milliliter]) and iohexol (Omnipaque 350; GE healthcare, Carrigtohill, Co., Cork, Ireland, 350 mg I/mL). The data records showed that twenty-five patients underwent an intravenous power injection as a bolus of 1.2-mL/kg (kilogram) iodine-based contrast medium at 1 mL/second (s). Postcontrast CT images were performed after the complete injection of contrast medium. The axial sections of precontrast and postcontrast orbital CT images scanned along the transaxial direction with the sections parallel to the optic nerve along a line from the inferior border of the maxillary sinus to the middle part of the frontal region. A Hitachi CT scanner took the slice thickness for image viewing of axial images at 1.25 mm (mm) and other MDCT scanners at 2–4 mm. The coronal and sagittal sections of postcontrast orbital CT images were reformatted with 2–4 mm in slice thickness. The reconstruction matrix for MDCT scans of orbit was 512 × 512.

#### *2.3. Analysis of Images and Pathologic Diagnosis*

Two experienced radiologists (CHW and YYC) reanalyzed orbital lesions of the 25 patients on orbital CT images with axial, sagittal and/or coronal images together without knowledge of pathologic diagnosis of orbital tumors. The consensus from the two radiologists served as the final interpretation of images. If the two radiologists could not reach an agreement on any features from orbital CT scans, a third experienced radiologist (HCH) mediated the disagreement.

This study analyzed the following orbital CT features of each lesion or of each patient: maximum diameter of a lesion, lesion borders, homogeneity of CT density, a lesion involving extraocular muscle tendons, the lacrimal sac, lacrimal gland, preseptal space, extraconal, conal or intraconal orbital compartments, the optic nerve, infraorbital nerve, presence or absence of bone remodeling, single tumor or multiple lesions and single or bilateral orbital involvement, single or bilateral lacrimal gland involvement. This study also measured the mean values of the precontrast and postcontrast CT Hounsfield unit scales (CTHU) of each orbital lesion among the 25 patients.

A regular border of an orbital tumor on CT scans indicated the contour of a lesion from the surrounding tissue was smooth for more than 75% of the lesion. An irregular border of a lesion showed microlobulated, microangulated or indistinct contour from the surrounding tissue with ≥25% of the lesion. Lesion involvement indicated lesion infiltration, invasion or encasement on orbital CT scans.

This study measured CTHU for all 47 orbital lesions of 25 patients on both pre and postcontrast prebiopsy CT scans. The region of interest (ROI) maker in an oval shape was placed in the center of each lesion to cover 50% of the largest tumor area on CT axial images, avoiding the inclusion of bone and blood vessels (Figure 1). This study also calculated postcontrast CTHU divided by precontrast CTHU.

**Figure 1.** A 68-year-old man with immunoglobulin G4 (IgG4)-related orbital disease (IgG4-ROD) shows multiple tumors in bilateral orbital cavities. (**a**) Axial computed tomography (CT) scans show masses in bilateral lacrimal glands (short and large arrows). The mean value of CT Hounsfield unit scale (CTHU) is measured at the enlarged right lacrimal gland (short arrows) on a picture archiving and communication system monitor. The region of interest (ROI) marker in an oval shape is placed in the center of the mass (short arrows) to cover 50% of the largest tumor area. The mean value of precontrast CTHU is 53 and that of postcontrast CTHU is 103. Sinusitis is found in the left frontal sinus with mucus retention (black curved arrows); (**b**) Coronal postcontrast CT scan shows multiple masses or enlargement in various ophthalmic tissues with regular borders and homogeneous contrast enhancement in bilateral orbital cavities as follows: a mass at the extraconal compartment of the right orbital cavity (black arrow),the right lacrimal gland (black curved arrow), the left lacrimal gland (large white curve arrow), the right superior rectus muscle belly (small white curve arrow) and the right infraorbital nerve (white arrow). Sinusitis is noted in the right ethmoid sinus with mucus retention (arrowhead); (**c**) Pathologic specimen shows infiltration of many lymphoplasma cells and mild fibrosis (hematoxylin–eosin stain, original magnification ×200); (**d**) Immunostaining for immunoglobulin G (IgG)-expression shows many plasma cells are positive for IgG stains (original magnification ×200); (**e**) Immunostaining for IgG4-expression shows abundant IgG4-positive plasma cells have infiltrated the lesion. IgG4-postive/IgG-positive plasma cell ratio is more than 40%. There are more than 100 IgG4-positive plasma cells in one high-powered field (>100/HPF) (original magnification ×200).

An experienced pathologist (AFYL) with 29 years of experience in pathology diagnosis reviewed the pathologic and immunohistochemical sections of the specimens of the 25 patients to confirm pathologic results of OL and IgG4-ROD. The two main pathologic criteria of IgG4-ROD included (1) IgG4+/IgG+ ratio > 40%, and/or (2) IgG4+ cells > 10/high-powered field (HPF) in histopathologic examination [5,9,14].

The radiologist (WHY) integrated demographic data, patient symptoms, signs and past histories, prebiopsy serum IgG4 levels and IgG levels and CT imaging interpretations and the mean values of CTHU measurement results of the 25 patients to evaluate the differences in CT qualitative and quantitative features, serum IgG and IgG4 levels between OL and IgG4-ROD.

#### *2.4. Statistical Analysis*

This study used SPSS version 19.0 software (SPSS, Inc., Chicago, IL, USA) for data analysis. Specially, we applied the Mann–Whitney U test to compare continuous variables because of the small sample size and the χ2 or Fisher's exact test for categorical variables at the level of significance of *p* < 0.05. Receiver operating characteristic (ROC) curve analysis calculated the area under the ROC curve to identify diagnostic values of CTHU, serum IgG4 levels and serum IgG levels of IgG4-ROD. This study assessed the findings based on sensitivity, specificity and accuracy with a 95% confidence interval (95% CI).

#### **3. Results**

The median age (mean ± standard deviation [SD], range) of the selected 25 patients was 59 (58.20 ± 10.61, 32–78). The median age (mean ± SD, range) of 16 patients with OL was 60.5 (59.31 ± 9.20, 41–78) and that of 9 patients with IgG4-ROD was 58 (56.22 ± 13.11, 32–69) (*p* = 0.934, Mann–Whitney U test). Of the 25 patients, 17 (68%) were male and 8 (32%) were female. Twelve (12 or 71%) of the 17 male patients were OL and 5 (29%) were IgG4-ROD; four (50%) of 8 females were OL patients and 4 (50%) were IgG4-ROD (*p* = 0.3942, Fisher's exact test).

The 25 patients showed proptosis, palpable mass and/or eyelid swelling—none of the 25 patients suffered from orbital pain or tender palpable mass. Six (6 or 24%) of the 25 patients had malignancy histories. Five (83%) of the 6 patients with malignant histories had OL: one with renal cell carcinoma, one with prostatic cancer and soft palate follicular lymphoma, one with squamous cell carcinoma of the tongue, one with follicular lymphoma involving lung, neck lymph nodes and bone marrow and one with chronic lymphocytic leukemia. Only one (17%) of the 6 patients with malignant history was an IgG4-ROD patient who had ovarian cancer. Patient malignant histories of the two groups had no significant difference (*p* = 0.3644, Fisher's exact test).

A pathologist (AFYL) reviewed the pathologic sections of the 25 patients. The pathologic review concluded 13 patients with extranodal marginal zone lymphoma of mucosa-associated lymphoid tissue (MALT lymphoma), 1 with low-grade B cell lymphoma with plasmacytic differentiation, 1 with diffuse large B cell lymphoma, 1 with follicular lymphoma and 9 with IgG4-ROD. The histopathologic findings of the 9 patients with IgG4-ROD showed diffuse lymphoplasmacytic infiltration, IgG4-positive (IgG4+) plasma cells, IgG-positive (IgG+) plasma cells and various degree fibrosis. Seven (78%) of the nine IgG4-ROD patients showed IgG4+ cells > 100 cells/HPF and IgG4+/IgG+ ratio > 40% (Figure 1). Another 2 of the 9 IgG4-ROD patients (22%) had IgG4+ plasma cell < 50 cells/HPF and IgG4+/IgG+ ratio > 40%.

Furthermore, CT images indicated a total of 47 orbital tumors among the 25 patients: 27 lesions were OL and 20 lesions were IgG4-ROD. Of 47 orbital tumors, none appeared inside the eyeball.

Tables 1 and 2 summarize CT features of 47 tumors among the 25 patients, of which 16 had orbital lymphoma and 9 had IgG4-ROD.


**Table 1.** Computed tomography (CT) features of 47 tumors among the 25 patients with orbital lymphoma or immunoglobulin G4-related orbital disease (IgG4-ROD) on prebiopsy orbital CT scans.

*n* (%)—number (percentage); SD—standard deviation; @—Mann–Whitney U test; #—Fisher's exact test.

**Table 2.** CT features of the 25 patients with orbital lymphoma or immunoglobulin G4-related orbital disease (IgG4-ROD) on prebiopsy orbital CT scans.


*n* (%)—number (percentage); #—Fisher's exact test.

Specifically, of the 16 patients with OL, eleven (69%) had a solitary tumor in an orbital cavity or at eyelids, 1 (6%) had 2 tumors, 2 (13%) had 3 and 2 (13%) had 4. The other 9 out of the 25 patients had IgG4-ROD: 2 (22%) with 1 tumor; 5 (56%) with 2 tumors, 1 (11%) with 3 tumors and 1 (11%) with 5 tumors. CT features statistically significantly associated with IgG4-ROD included lesions with regular borders (*p* = 0.0069), multiple tumors (*p* = 0.0414), lacrimal gland involvement (*p* = 0.0085), lesions involving bilateral lacrimal glands and bilateral orbital cavities (*p* = 0.0022 and *p* = 0.0168, respectively, Figure 1).

In contrast, tumors involving the extraconal, conal or intraconal space, lacrimal sac, optic nerve, extraocular muscle tendon, infraorbital nerve, preseptal space and presence of sinusitis and bone remodeling were ineffectual to differentiate IgG4-ROD from orbital lymphoma (Figures 1 and 2; all *p* > 0.05, Fisher's exact test). Two IgG4-ROD patients and 6 OL patients had a solitary tumor involving the preseptal space (Figure 2).

**Figure 2.** A 66-year-old man with orbital lymphoma shows a painless solitary lesion at the right lower eyelid. (**a**) Axial precontrast CT image shows a small nodule with an irregular border and homogeneous density involves the preseptal space of the right lower eyelid (arrowhead). Precontrast CT Hounsfield unit scale (CTHU) of the nodule is 57; (**b**) Axial postcontrast CT image shows the nodule demonstrates homogeneous enhancement (arrowhead). Postcontrast CTHU of the nodule is 68. Serum IgG4 level of the patient is 44.3 mg/dL.

Table 3 shows the descriptive statistical prebiopsy values of precontrast CT Hounsfield unit scales (CTHU), postcontrast CTHU and postcontrast CTHU/precontrast CTHU ratios of 27 tumors of OL and 20 tumors of IgG4-ROD on prebiopsy CT scans.

Table 4 demonstrates descriptive statistical prebiopsy values of serum IgG4 levels, serum IgG levels and the ratios of serum IgG4 level/serum IgG level of the 16 patients with OL and the 9 patients with IgG4-ROD.

**Table 3.** Descriptive statistical prebiopsy values of precontrast CT Hounsfield unit scales (Pre HU), postcontrast CTHU (Post HU) and postcontrast CTHU/precontrast CTHU ratios (Post HU/Pre HU) of 27 tumors of orbital lymphoma and 20 tumors of IgG4-related orbital disease (IgG4-ROD) on prebiopsy orbital CT scans.


*n*—number; SD—standard deviation.

**Table 4.** The descriptive statistical prebiopsy values of serum IgG4 levels (serum IgG4), serum immunoglobulin G (IgG) levels (serum IgG) and the ratios of serum IgG4 level/serum IgG level (serum IgG4/IgG) of the 16 patients with orbital lymphoma and the 9 patients with IgG4-ROD.


*n*—number; mg/dL—milligram/deciliter; SD—standard deviation.

Figures 3 and 4 show the differences in postcontrast CTHU, postcontrast CTHU/precontrast CTHU ratios, serum IgG4 levels and serum IgG4 level/serum IgG level ratios were statistically significant between the two groups (all *p* < 0.05, Mann–Whitney U test).

Figure 3 shows that the areas under the ROC curve (AUC) of precontrast CTHU, postcontrast CTHU and the ratios of postcontrast CTHU/precontrast CTHU were 0.56 (95% CI: 0.393–0.727, *p* = 0.484), 0.766 (95% CI: 0.615–0.917, *p* = 0.002) and 0.670 (95% CI: 0.498–0.842, *p* = 0.048). According to Figure 4, the AUC for serum IgG4 levels, serum IgG levels and the ratios of serum IgG4/serum IgG were 0.847 (95% CI: 0.674–1.000, *p* = 0.005), 0.684 (95% CI: 0.455–0.913, *p* = 0.134) and 0.819 (95% CI: 0.639–1.000, *p* = 0.009), respectively.

Compared with above data, the AUC using postcontrast CTHU (= 0.766) and serum IgG4 levels (= 0.847) was moderately accurate for the diagnostic yield of IgG4-ROD because both AUC measures fell between 0.7 and 0.9. The largest Jordon index 0.528 [(sensitivity−[1−specificity] = 0.528)] suggested a cutoff value of postcontrast CTHU equal to 89.5. The sensitivity and specificity were 0.75 (95% CI: 0.509–0.913) and 0.778 (95% CI: 0.578–0.914), respectively. As to serum IgG4 level, the largest Jordon index (0.701) suggested a cutoff value equal to 132.5 mg/dL, which resulted in sensitivity of 0.889 (95% CI: 0.518–0.997) and specificity of 0.813 (95% CI: 0.544–0.960), respectively.

For patients with postcontrast CTHU ≥ 89.5 in at least one orbital nodule in the two groups, 7 (58%) were IgG4-ROD patients and 5 (42%) were OL patients (*p* = 0.0414, Fisher's exact test). For serum IgG4 levels ≥ 132.5 mg/dL, 8 (73%) were IgG4-ROD patients and 3 (27%) OL (*p* = 0.0021).

The postcontrast CTHU and serum IgG4 levels for patients with a solitary orbital tumor in the two groups of OL and IgG4-ROD patients were as follows: higher postcontrast CTHU (≥89.5) in 3 OL patients and 1 IgG4-ROD case; lower postcontrast CTHU (<89.5) in 8 OL and 1 IgG4-ROD; and lower serum IgG4 levels (<132.5) in all 11 OL patients and 1 IgG4-ROD case with higher CTHU. In addition, one IgG4-ROD patient with a lower postcontrast CTHU showed a higher serum IgG4 level ≥ 132.5.

**Figure 3.** Box-and-whisker plots and receiver operating characteristic (ROC) curve analysis of precontrast computed tomography (CT) Hounsfield unit scale (CTHU), postcontrast CTHU, postcontrast CTHU/precontrast CTHU ratios between orbital lymphoma (OL) and immunoglobulin G4-related orbital disease (IgG4-ROD). Bars = medians. (**a**) Precontrast CTHU (precontrast HU) shows a nonsignificant difference between the two groups (*p* = 0.484, Mann–Whitney U test); (**b**) Postcontrast CTHU (postcontrast HU) shows a significant difference between the two groups (*p* = 0.002, Mann–Whitney U test); (**c**) Postcontrast CTHU/precontrast CTHU ratios (postcontrast HU/precontrast HU) show a significant difference between the two groups (*p* = 0.048, Mann–Whitney U test); (**d**) Areas under the ROC curve (AUC) of precontrast CTHU (precontrast HU), postcontast CTHU (postcontrast HU) and postcontrast CTHU/precontrast CTHU ratios (postcontrast HU/precontrast HU) are 0.56 (95% CI: 0.393–0.727, *p* = 0.484), 0.766 (95% CI: 0.615–0.917, *p* = 0.002) and 0.670 (95% CI: 0.498–0.842, *p* = 0.048), respectively. —an outlier.

**Figure 4.** Box-and-whisker plots and receiver operating characteristic curve (ROC) analysis of serum immunoglobulin G4 (IgG4) levels, serum immunoglobulin G (IgG) levels and serum IgG4 level/serum IgG level ratios between orbital lymphoma (OL) and IgG4-related orbital disease (IgG4-ROD). Bars = medians. (**a**) Serum IgG4-levels demonstrate a significant difference between the two groups (*p* = 0.004, Mann–Whitney U test); (**b**) Serum IgG levels demonstrate a nonsignificant difference between the two groups (*p* = 0.136, Mann–Whitney U test); (**c**) Serum IgG4 level/IgG level ratios (serum IgG4 level/IgG level) demonstrate a significant difference between the two groups (*p* = 0.009, Mann–Whitney U test); (**d**) The areas under the ROC curve of serum IgG4 level (serum IgG4), serum IgG level (serum IgG) and serum IgG4 level/IgG level ratio (serum IgG4/IgG) are 0.847 (95% CI: 0.674 to 1.000, *p* = 0.005), 0.684 (95% CI: 0.455 to 0.913, *p* = 0.134) and 0.819 (95% CI: 0.639 to 1.000, *p* = 0.009), respectively. \* and —outliers.

The postcontrast CTHU and serum IgG4 level for patients with multiple orbital lesions in the two groups were as follows: lower postcontrast CTHU (<89.5) were noted in 3 OL patients and in 1 IgG4-ROD case; higher postcontrast CTHU (≥89.5) were found in 2 OL patients and in 6 IgG4-ROD cases; higher serum IgG4 levels (≥132.5 mg/dL) were found in 3 OL patients and in 7 IgG4-ROD cases; lower serum IgG4 levels (<132.5 mg/dL) appeared in 2 OL patients. Two of the three OL patients with lower postcontrast CTHU showed OL involving bilateral lacrimal glands, who had different serum IgG4 levels: 51.3 and 339.5, respectively. Concurrent higher postcontrast CTHU and a higher serum IgG4 level were found in 1 OL patient (1/5, 20%) with multiple tumors in the left orbital cavity and in 6 IgG4-ROD cases with tumors involving bilateral lacrimal glands. Lower postcontrast CTHU and higher serum IgG4 level were noted in only one IgG4-ROD (1/7, 14%) case, who showed tumors mainly in intraconal spaces of bilateral orbits.

If this study used "lesions with bilateral lacrimal gland involvement" (the most significant qualitative CT feature in statistics, *p* = 0.0022), "bilateral lacrimal gland involvement and a higher serum IgG4 level (≥132.5 mg/dL) (*p* = 0.0005)" or "bilateral lacrimal gland involvement and higher postcontrast CTHU (≥89.5, quantitative CT feature) (*p* = 0.00047)" or "higher postcontrast CTHU and a higher serum IgG4 level" (*p* = 0.0029) as helpful test tools for diagnosis of IgG4-ROD (Table 5), sensitivity, specificity and accuracy of the first test (Test 1), the second (Test 2), the third (Test 3) and the latest one (Test 4) were as follows (Table 5): 0.78 (95% CI: 0.3999–0.972), 0.88 (95% CI: 0.617–0.985) and 0.84 (95% CI: 0.639–0.955) for Test 1; 0.78 (95% CI: 0.3999–0.972), 0.94 (95% CI: 0.698–0.998) and 0.88 (95% CI: 0.688–0.975) for Test 2; 0.67 (95% CI: 0.299–0.925), 1 (95% CI: 0.794–1) and 0.88 (95% CI: 0.688–0.975) for Test 3; 0.67 (95% CI: 0.299–0.925), 0.94 (95% CI: 0.698–0.998) and 0.84 (95% CI: 0.639–0.955) for Test 4. In Table 5, Test 3 had 100% of positive predictive value (PPV). Test 1 & Test 2 had the highest negative predictive value (NPV) 0.88.


**Table 5.** Contingency table of four helpful testing tools for diagnosis of IgG4-related orbital disease (IgG4-ROD).

*n*, patient number; Test 1 to Test 4 represent four helpful tools for diagnosis of IgG4-ROD; Test 1, orbital lesions with bilateral lacrimal gland involvement on CT scans; Test 2, orbital lesions with bilateral lacrimal gland involvement on CT scans and a higher serum IgG4 level (≥132.5 mg/dL [milligrams per deciliter]); Test 3, orbital lesions with bilateral lacrimal gland involvement and higher postcontrast CTHU (CT Hounsfield unit scales ≥ 89.5) on CT scans; Test 4, orbital lesions with higher postcontrast CTHU (≥89.5) and a higher serum IgG4 level (≥132.5 mg/dL); (95% CI), (95% confidence interval); \*—not shown in the statistics operation; PPV—positive predictive value—NPV—negative predictive value.
