**4. Discussion**

The IgG4-related disease can result in fibroinflammatory lesions at nearly any anatomic site [14]. OL is malignant and needs radiotherapy, chemotherapy and/or operation [4,7,8]. IgG4-ROD is benign and approximately 90% of patients respond well to steroid treatment [5]. Both of malignant OL and benign IgG4-ROD are lymphoplasmacytic infiltrated mass-like lesions, which make clinicians difficult to differentiate from each other [6]. This study showed that lesions with regular borders, multiple tumors, lacrimal gland involvement, simultaneous involvement of bilateral lacrimal glands and bilateral orbital cavities and higher medians of postcontrast CTHU and serum IgG4 levels were significantly related to IgG4-ROD (all *p* < 0.05). Postcontrast CTHU ≥ 89.5 showed 0.75 sensitivity and 0.778 specificity with the AUC = 0.766 (95% CI: 0.615–0.917, *p* = 0.002); serum IgG4 levels ≥ 132.5 mg/dL had 0.889 sensitivity and 0.813 specificity, with the AUC = 0.847 (95% CI: 0.674–1.000, *p* = 0.005, Figures 3 and 4). A lesion with regular borders is most likely to be a slow growing benign mass or less likely to be an indolent malignant tumor [1]. IgG4-ROD being benign usually presented as lesions with regular borders in this study.

Serum IgG4 levels account for 3% to 6% total amount of serum IgG levels [15]. Hamano et al. [16] reported a cutoff value of 135 mg/dL to differentiate autoimmune pancreatitis from pancreatic cancer with a high sensitivity (95%), specificity (97%) and accuracy (97%). This study identified a cutoff value 132.5 mg/dL (close to 135 mg/dL) to distinguish IgG4-ROD from OL at diagnostic accuracy (AUC) of 84.7% with 88.9% sensitivity and 81.3% specificity.

However, approximately 40–50% of patients with biopsy-proven IgG4-related disease have normal serum IgG4 concentrations [10,14,17]. In our study, normal serum IgG4 (<132.5 mg/dL) occurred in 13 (81%) of 16 OL patients and one IgG4-ROD patient (1/9, 11%), who had a solitary orbital lesion. There may be several reasons to explain why in our study there was a lower percentage of IgG4-ROD with normal serum IgG4 levels: first, our study was a small sample research, which may have selection bias; second, serum IgG4 levels may vary according to the specific organ involved [10]; finally, elevated serum IgG4 levels represent a subtype of IgG4-related disease with more inflammatory features and worsening disease activity [17]. Our IgG4-ROD patients (8/9, 89%) could be developing an active IgG4-related disease with elevated serum IgG4 concentrations.

Patient's age, standard imaging features and localizing orbital lesions to intraconal, conal or extraconal compartments help limit the differential diagnosis [18]. Our study showed no significant difference in median ages between patients with OL and IgG4-ROD (*p* = 0.934, Mann–Whitney U test). Lesions with regular borders, multiple orbital tumors, lacrimal gland involvement, lesions simultaneously involving bilateral lacrimal glands and bilateral orbital cavities and higher postcontrast CTHU (≥89.5) on orbital CT scans were significantly associated with IgG4-ROD (all *p*<0.05). The difference in extraconal, conal and intraconal compartments of orbital lesions between OL and IgG4-ROD groups was not statistically significant (*p* = 0.4813). In addition to CT, magnetic resonance imaging (MRI) also helps in further diagnostic workup of orbital tumors and provides ocular anatomy for lesions involvement, perineural spread and intracranial extension [12]. Both retinoblastomas typically found in children and uveal melanomas in adults appear in the globe. Retinoblastoma is slightly hyperintense on T1 weighted MRI (T1WI) and very hypointense relative to vitreous on T2-weighted MRI (T2WI) and well contrast enhancement on postcontrast CT and contrast-enhanced (CE) MRI [12,18]. Ninety percent of retinoblastomas demonstrate calcifications on precontrast CT scans [18]. Melanomas with melanin show characteristic hyperintensity on T1WI and hypointensity on T2WI [12,18]. For intraconal orbital tumors, gliomas common among children result in fusiform enlargement of the optic nerve on axial CT and MRI [12,18]. In contrast, meningiomas, commonly seen in the 5th decade of life, classically show the contrast-enhancing tumor with a "tram-tract" configuration alongside the optic nerve on axial postcontrast CT or CE MRI [12]. The most common benign orbital tumor in adults is a cavernous hemangioma, which typically demonstrates a well-defined dense unilateral orbital intraconal mass with intra-tumoral calcifications on precontrast CT scans and MRI. The enhancement spread pattern on a dynamic postcontrast CT and dynamic CE T1WI can help to distinguish between cavernous hemangioma and schwannoma [18,19]. Cavernous hemangiomas show initial patchy enhancement on arterial phase, but schwannomas start a wide area of enhancement. The most common congenital orbital nodules are dermoids, which usually show a well outlined round or oval tumor with a capsule and low density or fat contents in the extraconal space on CT scans or MRI [18]. Due to fat contents, dermoids typically show hyperintensity on T1WI, hyperintensity on T2WI and hypointensity on short tau inversion recovery MRI (STIR) [18]. Benign mixed tumor of lacrimal gland usually seen in middle-aged patients demonstrates a well-circumscribed round or oval tumor with homogeneous enhancement on postcontrast CT and CE MRI [12]. Malignant epithelial lacrimal gland tumors show a mass with a well- or poor-defined margin with associated bony remodeling or destruction in 70% cases on CT scans [18].

Multiple or multicompartmental orbital masses include venolymphatic malformations (VLM), rhabdomyosarcoma (RMS), plexiform neurofibroma, thyroid ophthalmopathy (TO), orbital pseudotumor (OP), lymphoma, metastases and IgG4-ROD [18]. The first three types of masses are common among children; the last five, among adults [3,5,18]. VLM usually appears poorly defined, lobulated and multiloculated lesions with various signal intensity on T1WI and T2WI [18]. VLM may demonstrate fluid–fluid level on MRI, which is highly suggestive of the diagnosis of VLM [12]. VLM, RMS and plexiform neurofibroma may have similar findings on CT and MRI [18]. TO causes enlarged

bilateral myositis of the extraocular muscles, often involves medial and inferior rectus muscles with sparing tendinous insertions on CT and MRI [12] and is related to elevated thyroid-stimulating hormone level [18]. OP, IgG4-ROD and OL show similar MRI features on conventional sequences, which are hypointense on T1WI and T2WI and well contrast enhancement on postcontrast T1WI. Furthermore, diffusion-weighted imaging (DWI) with apparent diffusion coefficient (ADC) mapping can help to differentiate between benign and malignant orbital lesions [1,18]. Sepahdari et al. have reported that an ADC value < 1.0 <sup>×</sup> 10−<sup>3</sup> mm2/ sec and an ADC ratio < 1.2 are optimal for predicting orbital malignant tumors [1,18]. Prior studies used ADC value <sup>&</sup>lt; 1.0 <sup>×</sup> <sup>10</sup>−<sup>3</sup> mm2/ sec and ADC ratio <sup>&</sup>lt; 1.2 to differentiate orbital lymphoma from benign OP and IgG4-ROD with more than 95% accuracy [1,18]. However, ADC values and ratios cannot differentiate OP from IgG4-ROD because the two disease have similar these values [1,18]. OP manifests with the most common acute unilateral painful mass in adults, which assist in differentiating OP from TO, OL and IgG4-ROD. Pain is uncommon in TO, OL and IgG4-ROD [12,18]. None of 25 patients in this study suffered from orbital pain or tender palpable mass, either. Consistent with the result of our research, Fujita et al. have reported that IgG4-ROD commonly presents involving bilateral lacrimal glands [15], which can distinguish OL and OP from IgG4-ROD. A clinician can suggest the diagnosis of orbital metastasis only when clinically primary malignancy is known [18].

An effective clinical diagnosis or appropriate disease classification for IgG4-related disease needs the integration of clinical findings, radiological features and serologic or pathologic data [12]. None of our 25 patients suffered from painful orbital lesions. Four tests for IgG4-ROD diagnosis used in this study included serologic serum IgG4 levels, radiological CT qualitative (lesions with bilateral lacrimal gland involvement) and/or quantitative features (postcontrast CTHU ≥ 89.5). Of the four tests, Test 2 (lesions with bilateral lacrimal gland involvement and a higher serum IgG4 level [≥132.5 mg/dL]) with the highest sensitivity (78%), a higher specificity (94%), a higher PPV (88%), the highest NPV (88%) and the highest accuracy (88%) could be the better prebiopsy test to distinguish IgG4-ROD from OL.

This study had only two IgG4-ROD patients with a solitary tumor, which was a small sample and lacked specific CT features. Prebiopsy diagnosis of a solitary IgG4-ROD could depend on a painless orbital mass, postcontrast CTHU ≥ 89.5 and serum IgG4 level ≥ 132.5 mg/dL. Tissue proof is an ultimate diagnostic way. However, biopsy is not always suitable for orbital lesions. The best medical option for a benign mixed tumor or malignant mass of the lacrimal gland may be excision en bloc without biopsy once clinical and imaging diagnosis. However, incomplete excision or ruptures of neoplasms via biopsy may result in tumor recurrence, malignant transformation of a mixed tumor and poor prognosis [12]. In clinical practice, some IgG4-ROD patients may be at high risk for biopsy and/or refuse biopsy. However, once these patients meet possible diagnosis of IgG4-ROD [20], systemic steroid treatment may be a good alternative. The criteria for possible IgG4-ROD diagnosis include elevated serum IgG4 (≥135 mg/dL), enlargement of the lacrimal gland or masses, enlargement or hypertrophic lesions in various orbital tissues [20]. Clinicians could forgo further biopsy if such patients respond well to glucocorticoids within weeks, such as reductions in the size of tumors, improvements of symptoms and a significant decrease in serum IgG4 [20]. Alternative non-vital organ or lip biopsy may be an acceptable option [21].

Sato et al. [22] reported that 17 (81%) of 21 patients with IgG4-ROD had involvement of the lacrimal glands and 13 (70.6%) of 17 cases showed bilateral lacrimal gland swellings. Our study also showed that seven (78%) of nine patients with IgG4-ROD had bilateral lacrimal gland involvement, which could distinguish IgG4-ROD from OL (*p* = 0.0022). Neither Sato et al. nor this study had patients with IgG4-ROD originating from conjunctival or subconjunctival tissue. However, prior research suggested that IgG4-ROD can also develop at conjunctival tissue [23,24].

This study has several limitations. First, this retrospective research had a small sample of patients. Second, the patients in this study received different brands of contrast agents and CT machine, which could produce potential bias in measurement of CTHU. Last, incomplete data records ruled out the

possibility of deciphering a detailed correlation between CT features and clinical presentations of OL and IgG4 ROD.

#### **5. Conclusions**

This study compared prebiopsy precontrast and postcontrast CT features, serum IgG4 and serum IgG levels of IgG4-ROD with those of OL. The key findings showed that IgG4-ROD had high correlation with the presence of lesions with regular borders, presence of multiple lesions, lesions involving the lacrimal gland, both lacrimal glands and bilateral orbital cavities on CT scans, higher values of postcontrast CTHU, postcontrast CTHU/precontrast CTHU ratios, serum IgG4 levels and serum IgG4/IgG level ratios (all *p* < 0.05). For diagnosis of IgG4-ROD, postcontrast CTHU ≥ 89.5 and serum IgG4 level ≥ 132.5 mg/dL provided moderate diagnostic accuracy, AUC = 0.776 and 0.847, respectively, which were higher than those of postcontrast HU/precontrast HU and serum IgG4/IgG level ratio. The special CT features and elevated serum IgG4 levels could help differentiate IgG4-ROD from OL. Prebiopsy diagnosis of the uncommon solitary type of IgG4-ROD could also depend on a painless orbital mass and elevated CTHU ≥ 89.5 and serum IgG4 level ≥ 132.5 mg/dL.

**Author Contributions:** Conceptualization, W.-H.Y. and A.F.-Y.L.; methodology, W.-H.Y. and A.F.-Y.L.; software, W.-H.Y.; validation, W.-H.Y. and A.F.-Y.L.; formal analysis, Y.-Y.C., C.-H.W., H.-C.H., A.F.-Y.L.; investigation, Y.-Y.C., C.-H.W., H.-C.H., A.F.-Y.L.; resources, S.-Y.Y.; data curation, S.-Y.Y., W.-H.Y.; writing—original draft preparation, W.-H.Y.; writing—review and editing, W.-H.Y.; visualization, W.-Y.G.; supervision, J.-F.L. and W.-Y.G.; project administration, W.-H.Y.; funding acquisition, W.-H.Y. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Acknowledgments:** We thank May Yuan for English language editing.

**Conflicts of Interest:** The authors declare no conflict of interests.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
