One of the major impacts of renal function in the evaluation of orbital pathology is the use of intravenous contrast agents for CT and MRI. The safety of intravenous contrast agents is highly dependent on renal function. Intravenous contrast (IVC) is the most common cause of iatrogenic acute kidney injury [16]. Contrast-induced nephrotoxicity (CIN) is uncommon in patients with normal kidney function, with an incidence of 1–2%. However, the incidence rises dramatically to as high as 25% in patients with chronic kidney disease [16]. CIN typically occurs within 3 days of contrast infusions with a peak serum creatinine at 3–5 days, usually returning to baseline in 7–10 days. Prevention of CIN mainly involves adequate precontrast hydration, either with intravenous isotonic saline or with unrestricted oral fluid intake [17]. Nonionic iodinated contrast material appears to be safer for the kidneys than ionic iodinated contrast material, at least when measured by creatinine clearance. As an aside, CIN from appropriately dosed gadolinium infusion is either nonexistent or exceedingly rare.

With regard to CT contrast agents, several issues must be stressed. First, because of the potential for seizure and vasculopathic complications with iodinated contrast, nonionic contrast agents are used in the vast majority of central nervous system imaging with CT, including the orbit [18]. Fatal reactions are rare, ranging between 1 in 20,000 and 40,000 examinations utilizing contrast material; of note, fatality rates are similar between high-osmolar ionic contrast and low-osmolar nonionic preparations. However, transient ischemic attacks and cerebrovascular infarctions during cerebral angiography occur much more frequently, at a rate of 1.6% in high-risk patients. Second, the seemingly age-old custom of inquiring about shellfish allergies as a possible harbinger of contrast allergy is unfounded as no reliable link has been identified. However, any history of a previous contrast-related reaction is important, as this increases the risk of subsequent reaction fivefold. Atopy (including asthma) also increases the risk of allergic reaction.

Intravenous corticosteroids have been recommended as a “premedication” for at-risk patients. It is important to note, however, that there is no unequivocal proof of short-term corticosteroid efficacy in preventing contrast-induced anaphylactoid reactions. Furthermore, adequate suppression of basophils and histamine production is typically delayed for 4–6 h after corticosteroid infusion, and such a delay may be untenable in emergency situations. One common algorithm in at-risk patients is an oral dosing of corticosteroids beginning at least 6 h prior to contrast infusion, with or without oral antihistamines. Low-osmolarity nonionic agents appear to have a lower incidence of allergic reaction. An excellent summary for management of a variety of contrast-induced reactions is available for download from the American College of Radiology (ACR) Manual on Contrast Media, version 9, 2013 [19].

Nephrogenic systemic fibrosis (NSF) is a scleroderma-like syndrome related to the infusion of gadolinium during MRI [20]. It is most frequently associated with renal failure, may be progressive, and may affect multiple organ systems. NSF is admittedly rare, and its incidence depends on the specific type of gadolinium compound used. As an example, gadodiamide has an NSF incidence of 1:2913 (0.03%), while gadopentetate dimeglumine has a much lower incidence of only 1:44,224 (0.002%) [20–22]. The overall incidence of NFS appears to be decreasing in the United States, likely because of the avoidance of higher-risk gadolinium preparations, more effective patient screening, and avoidance of use of gadolinium in high-risk patients, as well as more discriminant use of other MRI contrast materials.

NSF typically occurs in patients with the clinical triad of gadolinium infusion for MRI, renal failure (either chronic or acute), and a “pro-inflammatory state” (recent major surgery, endovascular injury, sepsis, malignancy, thrombosis, etc.) [23, 24]. In patients with significant renal failure, the incidence of NSF is estimated at about 4%, with 45% of affected patients suffering from severe disability and with a mortality of up to 31% [23, 24]. Of note, no effective therapy exists for treatment of NSF, and there is no evidence that pretreatment of patients with corticosteroids or pre−/post-gadolinium dialysis has any effect on the incidence and severity of NFS. Skin changes in NSF begin days to months after gadolinium infusion, with an average of 11.5–14 days [23–25].

Although NSF is rare, the severity and potential mortality of the syndrome requires cautious use of gadolinium in patients at higher risk for NSF [22, 23]. The US Food and Drug Administration (FDA) has published recommendations regarding gadolinium use, which are summarized in Table 32.2 [26].

The American College of Radiology has recently noted in its Manual on Contrast Media (version 10.2) that dose-dependent gadolinium deposition has been noted within brain tissue in patients who underwent serial contrast-enhanced MRIs [19]. Deposition mainly occurred in the dentate nucleus and globus pallidus, but histopathologic examinations have also revealed gadolinium in the frontal lobe white matter and in the prefrontal cortex [27, 28]. Importantly, the deposition can occur in patients with an intact blood–brain barrier and no evidence of hepatic or renal disease and has been documented in children undergoing serial MRI with contrast. Thus far, the clinical significance of these findings is unknown, and no adverse effects have been noted, but the FDA is actively investigating this phenomenon. In the meantime, clinicians should be aware of this issue. Gadolinium infusion during MRI frequently provides critical imaging information and often needs to be used in initial, routine MRI [27]. However, in cases where serial MRI is indicated in a child (e.g., monitoring for glioma progression), the utility of a non-contrast MRI should be discussed with the neuroradiologist. Some experts also recommend the use of higher stability, macrocyclic gadolinium preparations, which have lower deposition rates [27, 28].

A significant amount of data has been published since 2001 on the subject of radiation exposure from computed tomography (CT) [29–32], resulting in a broad spectrum of interpretation by the media and in the literature [33, 34]. This, in turn, has led to appropriate concern among patients and the parents of young patients as to the potential cancer risk from CT. This section reviews the available data and summarizes the conclusions and recommendation of our radiology colleagues. Of note, most of the concerns regarding radiation effects of CT pertain to children (especially young children) as compared to adults [34].

First, it is critical to remember that CT is an extremely powerful diagnostic tool that is readily available and is appropriate to use when clinically justified, even in children [35]. As a blanket statement, the risk of potential future cancer from a single CT is far less than the diagnostic benefit, especially in emergency situations [36, 37]. CT requires the use of a certain amount of ionizing radiation  – there is simply no way around the basic physics of the technology. However, the radiation dosage of a diagnostic CT can be adjusted to lower levels without compromising diagnostic quality – the ALARA concept (as low as reasonably achievable).

Until recently, the actual risk of future cancer from low-level radiation exposure was not definitively quantified by any large study [36]. The majority of published conclusions relied on extrapolation from atomic bomb victims where there was whole-body exposure or from mathematical models [34, 38]. However, since 2012, two major studies have been published which quantify CT radiation risk and have reached some sobering conclusions [39, 40].

The study of ionizing radiation effects on the body is complex and beyond the scope of this chapter. Briefly, however, radiation affects specific tissues in different ways. As an example, studies on atomic bomb survivors concluded that the lung is more sensitive to the carcinogenic effects of radiation exposure than the liver, skeletal muscle, or skin [29, 37]. Furthermore, the dose of radiation from a CT study that strikes the body (“entrance dose”) is not equivalent to the absorbed radiation by the body [39]. It is difficult to accurately quantify the risk of future malignancy from low-dose radiation exposure because low-dose radiation is a weak carcinogen [36, 39]. However, it is important to note that three expert panels, including the National Academy of Sciences, have concluded that there is a potential cancer risk even with low-level exposure and that this risk rises with increasing exposure [37]. In addition, the concept of “radiation effect” is critical in understanding the radiation risk from CT. Put simply, two types of radiation effects exist [34]. Deterministic effects are well known to the orbitologist and are typified by high-dose radiation to the orbit (or any anatomic structure) over a short period of time, usually for the treatment of malignancy, resulting in myriad side effects (e.g., keratopathy, retinopathy, optic neuropathy, etc.). Stochastic effects are secondary to low doses of ionizing radiation absorbed over a long period of time, as seen in serial imaging, be it CT, mammography, etc. Although the side effects from such stochastic radiation are not immediately obvious, they do accumulate over time and are definitely more of an issue in the pediatric, as compared to the adult population.

The risk of radiation exposure is significantly different in children than in adults for multiple reasons [37]. First, children are more sensitive to the effects of radiation exposure than adults because of active tissue growth and development, with an estimated tenfold overall increased neoplastic potential [30]. Second, because of their age and statistically longer life span, children have more time in the ensuing decades to manifest any ill effects of radiation exposure; put simply, it is far more likely to see a radiation effect that takes 40 years to manifest in someone who is 10 years old than in someone who is 50 years old although, until recently, the magnitude of this risk from low-level exposure was difficult to accurately quantify. Third, unless the baseline radiation exposure parameters are adjusted for children, an equivalent dose has a higher effect on the smaller cross-sectional anatomic area of a child than of an adult; even with such adjustment, the final radiation dose effect to a child may be up to 50% higher than in an adult [41, 42]. Fourth and perhaps most pertinent to the reader, the pediatric brain is especially sensitive to radiation effects [39], and at least a portion of the brain is unavoidably imaged during an orbital study. Finally, the ubiquity of CT in medicine and expanding indications for CT has led to ever-increasing utilization.

The use of diagnostic CT in the United States has risen by more than 20-fold, from 2 million studies annually in 1980 to almost 80 million in 2009 [31, 32]. Seven million CT studies are performed on children annually in the United States, and this number increases an average of 10% per year. Of note, the majority of brain CTs performed in children is for head trauma [30, 40, 43–45]. CT also accounts for a disproportionately high amount of total radiation exposure in medicine. A study from one academic radiology department published by Mettler and colleagues in 2000 concluded that although CT accounted for only 11% of procedures using ionizing radiation , 67% of all ionizing radiation was attributable to this modality [45].

Perhaps most importantly, CT of one organ system can be performed using a variety of techniques with a wide range of radiation exposure. This concept is extremely important to keep in mind for pediatric CT. As noted in a review by Brody and Frush, increasing the dose of radiation in an effort to improve image (and presumably diagnostic) quality has its limits, and after a certain threshold, additional radiation simply has no effect on improved diagnostic accuracy [37]. Therefore, lower doses of ionizing radiation in pediatric CT are preferable and do not necessarily compromise the quality of the study. The ALARA concept has markedly reduced radiation exposure in children from CT studies, by 50–90% since about 2001, and these techniques continue to undergo refinement [30, 37, 40, 46].

The results of two recent large studies on radiation exposure from CT in children are worth summarizing in detail. Pearce and colleagues analyzed a national health-care database in the United Kingdom and identified just over 175,000 pediatric patients that met the inclusion criteria of the study [39]. Of note, of 283,919 CT scans performed, the majority (64%) were of the head, an area of particular importance to the reader. The authors of the study estimated that the average total dose from one brain CT in a child ranged between 28 and 44 mGy depending on age and gender (Table 32.3) [39]. The rates of leukemia and brain tumors in patients undergoing CT were analyzed based on dose. A significantly increased risk of both maladies was found in patients who received certain threshold cumulative radiation doses. Specifically, the relative risk (RR) of leukemia in patients with a cumulative dose of at least 30 mGy (mean 51.13 mGy) was 3.18 compared to those children who received less than 5 mGy. With regard to brain tumors, the RR was 3.32 in patients with a cumulative dose of 50 mGy (mean 104.16 mGy). The authors estimated that for children younger than 15 years, these threshold cumulative doses amounted to five to ten head CTs for a 50 mGy bone marrow dose and two to three head CTs for a 60 mGy brain dose. Put another way, the authors concluded that one head CT before the age of 10 years would result in one excess case of leukemia and one excess brain tumor per 10,000 patients. Of note, multiple studies have concluded that younger children, especially those <5 years of age, are particularly sensitive to ionizing radiation [39, 40]. Admirably, Pearce et al. also pointed out that, first, pediatric leukemia and brain tumors are relatively rare events and that tripling this risk from CT radiation exposure still makes these entities rare events and, second, the increased lifetime cancer risk from CT is “very small” compared with the natural occurrence of malignancy in the general population which may make the benefits of CT more important “providing it is clinically justified.” Herein lies the crux of the discussion: while CT is a powerful and lifesaving technology, with a now-proven underlying cancer risk, it therefore should be used only when appropriate and then very judiciously.

In the second large study, Mathews and colleagues compared 680,211 pediatric patients who underwent CT with 10,939,680 who did not [40]. Mathews et al.’s study echoed the findings of Pearce et al. regarding subsequent leukemia development, providing independent confirmation of the previous study’s findings; a cumulative dose of 30 mGy resulted in a relative risk of 3.18. Brain CT was associated with an excess incidence rate for brain cancer of 2.97/100,000 person-years. Put another way, the authors estimated 1 excess brain malignancy per 4000 brain CTs. As already noted, the authors also confirmed that children <5 years of age had the highest relative risk. However, even older children and adolescents showed an increased cancer risk following CT. Serial CT also increased the risk, presumably due to increased cumulative radiation dosages. Of particular interest to the reader, facial CT was associated with only a borderline increased RR of subsequent malignancy (IRR = 1.14). Since an orbital study is essentially a subseries of a standard maxillofacial protocol, it is reasonable to extrapolate this facial CT finding for orbital scans. In other words, based on the findings of Mathews et al., an orbital series should have a lower risk for subsequent brain or hematopoietic tumor development than a standard brain CT. However, this statement must be tempered by the varying protocols used at different health-care facilities, some of which may extend a standard orbital series more cephalad, thereby exposing more of the brain to radiation, as well as the fact that in many (if not most) orbital injury cases, the head is scanned along with the orbits as a precaution.

Mathews and colleagues also brought up some important limitations to their study (which do not change, however, the prevailing opinion among radiologists that even low-dose radiation exposure increases cancer risk, albeit by a very small amount). First, as already noted, the authors point out that CT dosing protocols, especially for pediatric scans, have undergone a profound change since 2001 and that the current cancer risk from CT radiation may be smaller than calculated from older data. Second, the possibility of reverse causation may explain some of the increased brain tumor risk seen from head CT. In other words, symptomatology may have prompted initial brain imaging, which failed to diagnose an early, small lesion that went on to grow until the diagnosis was made several years later; thus, the undiagnosed tumor prompted the initial CT, and therefore the CT did not cause the tumor. Of note, the authors still found an increased solid and hematopoietic cancer risk following CT exposure, even when all brain tumors were excluded from analysis.

What, then, is the role of CT in pediatric orbital disease? First, as already stated, it is important to recognize that CT remains a valid and powerful tool in the pediatric population. In cases of head and orbital trauma, CT remains the imaging modality of choice because of its ready availability, rapid acquisition time, excellent resolution of bony and soft tissue orbital anatomy, and reasonable resolution of brain anatomy, especially in cases of acute intracranial hemorrhage. That said, it is also important to remember that a pediatric brain CT carries an increased risk of brain malignancy when compared to facial (i.e., orbital) CT based on the results of Mathews et al.’s study and a brain CT should not be a knee-jerk response to every case of pediatric orbital trauma.

The authors recommend the following guidelines for the judicious and appropriate use of orbital CT, much of it gleaned from the recommendations of the American College of Radiology :

Table 32.4 from the American College of Radiology provides an excellent summary of various orbital imaging options and their relative radiation risk for a child presenting with exophthalmos [49]. Note that MRA carries a safer rating than CTA and does not require ionizing radiation . It is also important to note that a plain X-ray of the orbit carries the worst rating because it provides essentially no clinical information despite exposure to (albeit low) doses of ionizing radiation .

Imaging is helpful in delineating and characterizing globe abnormalities, which include anophthalmia (Fig. 32.3), microphthalmia (Figs. 32.4 and 32.5), macrophthalmia (Fig. 32.6), colobomas (Figs. 32.7 and 32.8), buphthalmos, and staphylomas. In ophthalmology, colobomas refer to defects in ocular structures that result from incomplete closure of the fetal (embryonic) fissure. Colobomas are common developmental malformations and usually are bilateral [50]. In cases of coloboma, the optic nerve head may be small and slit-like (Fig. 32.7), or there may be a large, funnel-shaped staphylomatous excavation with the optic disc at the apex of the defect. The latter type of coloboma is also referred to as morning glory syndrome. Large colobomas are easily detected (Fig. 32.8), but even small defects in the optic nerve heads can be shown on images (Fig. 32.7) when a high-resolution, thin-section T2 sequence is utilized or with a particularly specific sequence such as HASTE (half-Fourier acquisition single-shot turbo spin echo) which shows excellent contrast between fluid-filled and soft tissue structures (Fig. 32.9). Microphthalmos with cyst represents a severe colobomatous malformation when the inner layer of retina proliferates and herniates through the enclosed fetal fissure to form one or more cysts, which may be small (Fig. 32.10) or very large (Fig. 32.11).

Buphthalmos , or an enlarged eye (“ox eye”), is most commonly seen in primary congenital or early-onset infantile glaucoma; there is also an association with neurofibromatosis type 1, due to secondary glaucoma from neurofibromas. In addition, intraocular pathology, including tumor, can also lead to globe enlargement. In primary congenital glaucoma, there is resistance to normal outflow of aqueous humor due to an abnormally developed trabecular meshwork. This results in increased intraocular pressure, which causes enlargement of the globe and thinning of the sclera [51]. CT or MR demonstrates not only the macrophthalmia with an enlarged anterior chamber and flattened lens but also its effect on the bony orbit, which may have up to 15% increased volume (Fig. 32.6).

Minimal symmetrical hypoplasia of the optic nerves may be difficult to detect clinically as well as on images. However, if the optic hypoplasia is unilateral, making comparison to the normal side possible, or if the hypoplasia is moderate to severe, then it can be detected on images. In such cases, it is important to obtain thin, contiguous sections and to look at the nerves in more than one plane to avoid making an error due to volume averaging. Optic nerve hypoplasia may be an isolated anomaly or may be associated with other abnormalities, as in the case of septo-optic dysplasia (Fig. 32.12) [52]. Coronal T2-weighted MRI is usually the best technique for evaluation of optic nerve size.

In cryptophthalmos , there is failure of development of lid folds, and as a result, the eyes are hidden by continuous overlying skin (Fig. 32.13) [53, 54]. The underlying globes typically are abnormal, having large anterior colobomas with absence of lenses or abnormal anterior chambers (Fig. 32.13). Cryptophthalmos is part of a syndrome that includes anomalies of extremities with craniofacial and urogenital abnormalities.

Dermoid and epidermoid cysts are the most frequent developmental cysts to occur in the orbit and represent choristomas, which are malformations of tissue that normally are not found at a given anatomic site. These cysts result from sequestration of ectoderm in the wall of the orbit or within the lids by advancing midfacial and frontal waves of neural crest during embryogenesis and typically occur either superolaterally (frontozygomatic suture) or superomedially (frontoethmoidal suture). Dermoid cysts, which are more common than epidermoid cysts, are lined by keratinizing squamous epithelium and contain dermal appendages (sebaceous glands, hair follicles) and keratinaceous debris and fatty substances. Epidermoid cysts are lined by epidermis without adnexal structures and contain desquamated keratin. The slow enlargement of the cysts causes pressure erosion of the adjacent bone, resulting in flattening (Fig. 32.14), thinning, and scalloping of the bone (Fig. 32.15). Occasionally such a cyst may break through the bone and displace dura mater. These developmental cysts have well-defined margins unless they have ruptured and caused an inflammatory reaction (Fig. 32.16). Also, the cysts may be associated with tracts that extend into adjacent structures (e.g., temporalis fossa, cranial vault, paranasal sinuses).

The density on CT or the intensity on MRI of dermoid or epidermoid cysts ranges from that of cerebrospinal fluid to that of fat, and the lesions may be heterogeneous due to mixture of fat and other desquamated debris (Fig. 32.17) [55]. The heterogeneity of the signal coupled with pockets of hyperintensity on precontrast T1-weighted images helps to distinguish dermoid/epidermoid cysts from other orbital pathologies. Only a few orbital processes manifest this hyperintensity on precontrast T1-weighted images, and these exceptions are worth remembering, as they narrow the differential diagnosis (Table 32.4). Occasionally, there may be calcification in the capsule of a dermoid/epidermoid cyst. As just noted, identification of the lipid/protein content leads to the correct diagnosis, but most cysts tend to be isointense to extraocular muscle, and the diagnosis is often based on location and morphology. Only the cyst capsule shows enhancement with contrast. On DWI, most cysts show restricted diffusion – an important sign that reinforces the diagnosis of dermoid/epidermoid cyst (Fig. 32.18).

A developmental cyst should be differentiated from dacryocystocele (Fig. 32.19), which forms due to obstruction at the intranasal opening of nasolacrimal duct (valve of Hasner) and may be seen in a newborn. In such a case, one can identify a markedly dilated nasolacrimal duct (Fig. 32.19b); furthermore, the inferomedial location of a dacryocystocele is an uncommon site for dermoid/epidermoid cyst formation.

Orbital teratomas are choristomas (tissue at the wrong anatomic location) that are composed of tissue from all three germ cell layers – ectoderm, mesoderm, and endoderm – and are usually benign. These congenital tumors typically contain disorganized tissues but, rarely, may have fetal features. It is speculated that lack of normal response of tissue to inducers results in teratoma formation [56, 57]. Orbital teratomas generally present at birth with marked proptosis and then continue to grow rapidly [58]; they can also be demonstrated on fetal MRI. CT and MR demonstrate the teratomas as large multicystic masses that cause prominent expansion of the bony orbit (Fig. 32.20) [2]. Imaging studies are important in delineating teratomas and identifying their vascularity, determining whether there is an intracranial component and differentiating the teratomas from microphthalmos with a cyst, dermoid cyst, meningoencephalocele, and hemangioma or venolymphatic malformations located deep within the orbit.

Glial–neural hamartomas occur rarely in the orbit and, therefore, often are difficult to diagnose. However, they should be considered as a possible diagnosis when an orbital mass is discovered at birth or shortly thereafter and lacks the characteristics of the more frequent pediatric orbital tumors, such as a capillary hemangioma, lymphangioma, dermoid, or teratoma. The glial–neural hamartomas may be well defined or infiltrative (Figs. 32.21 and 32.22) [59]. The presence of other congenital abnormalities (Fig. 32.22d) such as an abnormal globe, optic nerve, or extraocular muscles may be helpful in reaching the correct diagnosis by indicating maldevelopment as an etiology. However, if orbital structures appear normally formed and the glial–neural hamartoma is poorly defined (Fig. 32.21), then such a glial–neural malformation may be difficult to differentiate from fibromatosis or a rhabdomyosarcoma.

Strabismus (ocular misalignment) occurs frequently in children and involves various clinical patterns. Although the most common type of strabismus is congenital, other conditions may mimic it, and these may be related to neurologic disease, which should be promptly diagnosed and treated [60]. In this regard, imaging plays an important role for it can demonstrate cranial base and intracranial lesions responsible for cranial nerve palsies that result in paralytic strabismus. MRI can reveal and characterize cavernous sinus lesions, such as cranial nerve schwannoma (Fig. 32.23), inflammation (Fig. 32.24), or aneurysm (Fig. 32.25). In addition, brain parenchymal traumatic, infectious, inflammatory, demyelinating, ischemic, and neoplastic lesions responsible for cranial nerve palsies are well demonstrated by MRI even when they are very small and located within the brainstem [61].

The excellent morphologic detail of the orbital structures provided by CT and MRI can also be helpful in the more complicated cases of ocular misalignment and fixation. Differential diagnosis can be provided between extraocular muscle absence (Fig. 32.26) [62], malfunction, and actual restriction due to fibrosis (Fig. 32.27) or fibrotic mass (Fig. 32.28). If there are complications after strabismus surgery (e.g., slipped muscle, excessive scarring) (Fig. 32.29), imaging can be utilized to assess the problem and to plan further management.