MR Imaging of Laryngeal and Hypopharyngeal Cancer

Laryngeal cancer represents about 1% to 2% of cancers worldwide and, although the incidence is decreasing in some countries, the global incidence is increasing.

MR Imaging of Laryngeal and Hypopharyngeal Cancer

Introduction

Laryngeal cancer represents about 1% to 2% of cancers worldwide and, although the incidence is decreasing in some countries, the global incidence is increasing. The highest incidence is seen in Europe followed by the Americas, whereas the ratio between death and incidence is highest in Africa. The worldwide incidence of hypopharyngeal cancer is about 0.5% to 1% with a well-documented increasing incidence in women in several countries. Over 95% of laryngeal and hypopharyngeal cancers are squamous cell carcinomas (SCCs) caused mainly by tobacco and alcohol consumption and, in contrast to oropharyngeal cancer, infection with the human papillomavirus does not play an important role.

Key Anatomic Considerations and Implications for Patterns of Submucosal Tumor Spread

Accurate radiological assessment of submucosal tumor spread plays a key role in the T classification of laryngeal and hypopharyngeal cancers and in treatment planning, especially for transoral laser excision, open partial laryngectomy, and highly focused radiotherapy. 

Paraglottic Space

The paraglottic space plays an important role in the T classification of laryngeal cancers because if it is invaded, SCC is classified as T3 according to the AJCC/UICC guidelines. Whether neoplastic involvement of the paraglottic space influences the outcome after radiotherapy is still a matter of debate. However, invasion of the posterior paraglottic space and of the thyrocricoarytenoid space (TCAS, see below) is associated with a poorer outcome in terms of overall survival, disease-free survival, and locoregional control after radiotherapy and open partial laryngectomy. Furthermore, if the posterior paraglottic space is invaded, transoral laser surgery is contraindicated.

The paired and symmetric paraglottic space found in the supraglottic and glottic region mainly contains adipose tissue, loose elastic and collagen tissue, and blood vessels. Laterally, the paraglottic space is bordered by the thyrohyoid membrane, thyroid cartilage, and cricothyroid membrane, medially by the quadrangular membrane, laryngeal ventricle, aryepiglottic muscle, and thyroarytenoid muscle (which forms the bulk of the true vocal cords), posteriorly by the mucosa of the piriform sinuses and inferiorly by the conus elasticus. In the posterosuperior supraglottic region, there is no boundary between the paraglottic space and the pre-epiglottic space. However, in the posteroinferior supraglottic region, the thyroglottic ligament (a fibrous septum which fans out from the vocal ligament to the thyroid cartilage) separates the two spaces. At the glottic level, the paraglottic space can be divided into an anterior part (two-thirds) and a posterior part (one-third) by a line extending laterally from the vocal process of the arytenoid to the thyroid lamina. The posterior paraglottic space is currently considered as part of the TCAS and neoplastic invasion of TCAS is an important adverse prognostic factor in laryngeal cancer.

The paraglottic space is best assessed on axial and coronal images. It is easily identified on MR imaging because of its high fat content and, therefore, has a high signal intensity on T1 and T2. The normal paraglottic space may show some minor enhancement after intravenous administration of gadolinium chelates as it contains blood vessels. MR imaging is superior to CT for assessing the narrow paraglottic space at the vocal cord level.

The paraglottic space plays a pivotal role in submucosal tumor dissemination and it constitutes a “highway” for submucosal tumor spread. Invasion of the paraglottic space occurs whenever SCC arising in the laryngeal ventricle, false cords, or true vocal cords spreads laterally or whenever SCC of the piriform sinus spreads anteriorly. Key points of submucosal tumor spread along the paraglottic space include the following:

• Tumor spread in the paraglottic space typically occurs in a vertical direction (transglottic spread).
• Lateral tumor spread from the paraglottic space results in thyroid cartilage invasion.
• Lateral tumor spread from the paraglottic space through the cricothyroid ligament inferiorly and thyrohyoid membrane superiorly leads to invasion of extralaryngeal soft tissues.

Tumor spread from the posterior paraglottic space leads to early involvement of the piriform sinus, cricoarytenoid joint, and posterior thyroarytenoid muscle.

Pre-epiglottic Space

The pre-epiglottic space plays an important role in the T classification of laryngeal cancers: if the pre-epiglottic space is invaded, SCC is classified as T3 according to AJCC/UICC guidelines. Multiple authors have suggested that the degree of invasion of the pre-epiglottic space influences surgical options and to a moderate degree the likelihood of local recurrence after radio(chemo)therapy and open partial laryngectomy although others have questioned the impact of pre-epiglottic space invasion on local control after concurrent radiochemotherapy.

The pre-epiglottic space mainly contains fatty tissue, some loose connective tissue, and blood vessels. Anteriorly, it is bounded by the thyrohyoid membrane and thyroid cartilage, posteriorly by the epiglottis, cranially by the hyoepiglottic ligament, and caudally by the thyroepiglottic ligament. It is important to bear in mind that the pre-epiglottic space extends not only anteriorly to but also posterolaterally to the epiglottis and the posterolateral border of the pre-epiglottic space on axial images is located at about the anteroposterior midpoint of the thyroid lamina. The pre-epiglottic space is best seen on axial and sagittal images and it is easily identified on MR imaging and CT because of its high fat content. The normal pre-epiglottic space behaves like fatty tissue on MR imaging. Minor reticulated contrast enhancement of the pre-epiglottic space can be seen on high-resolution contrast-enhanced T1 because of the presence of blood vessels and loose connective tissue. Early invasion of the pre-epiglottic space occurs when SCC arising from the laryngeal surface of the epiglottis spreads in an anterior direction. Key points of submucosal tumor spread involving the pre-epiglottic space include the following:

• As there is no effective boundary posteriorsuperiorly between the pre-epiglottic space and paraglottic space, tumor spread from the pre-epiglottic space can result in paraglottic space invasion and vice versa.
• Anterolateral tumor spread from the pre-epiglottic space through the thyrohyoid membrane results in invasion of extralaryngeal soft tissues.

Laryngeal Cartilages

Invasion of the thyroid and cricoid cartilages influences the T classification of laryngeal and hypopharyngeal cancers. Laryngeal SCC with invasion of the inner cortex of the thyroid cartilage is classified as T3, whereas laryngeal SCC with invasion of the cricoid cartilage or invasion through the outer cortex of the thyroid cartilage is classified as T4a. In contrast, hypopharyngeal SCC invading the thyroid or cricoid cartilage is classified as T4a irrespective of the degree of cartilage invasion. In addition, invasion of the laryngeal cartilages affects prognosis after radiotherapy and influences the choice of surgery (total laryngectomy vs open partial laryngectomy or transoral laser microsurgery).

The thyroid, cricoid and arytenoid cartilages (with the exception of the vocal process of the arytenoids which contains elastic fibrocartilage) are composed of hyaline cartilage. Hyaline cartilage undergoes ossification with increasing age. Ossified cartilage corresponds histologically to bone: it has an inner and outer cortex and a marrow cavity with predominantly fatty tissue, some erythropoietic marrow and bone trabeculae. Although the normal ossification process of laryngeal cartilages tends to follow a predefined pattern and can result in complete ossification of the thyroid and cricoid cartilage, in most individuals, the central thyroid laminae and parts of the cricoid cartilage remain nonossified throughout life, and differences in ossification between right and left are common. Furthermore, asymmetric ossification can be equally seen in normal arytenoid cartilages.

The mechanism by which cancer invades laryngeal cartilages involves the following steps:

• An inflammatory phase (with edema and increased vascularization) within the cartilage located in immediate tumor vicinity and before actual tumor invasion.
• An osteoblastic phase with new bone formation.
• An osteoclastic phase in which newly formed bone is eroded and frank invasion by tumor cells occurs.

Both early neoplastic cartilage invasion and inflammatory changes in the laryngeal cartilages due to tumor vicinity may manifest with increased cartilage ossification (sclerosis) at CT. Therefore, the distinction between the normal mix of nonossified and ossified cartilage, sclerosis due to tumor invasion and sclerosis due to peritumoral inflammation is not possible with CT. Normal nonossified hyaline cartilage and ossified cartilage display characteristic signal intensities on MR imaging. Hyaline nonossified cartilage and the cortex of ossified cartilage are strongly hypointense on T1 and T2 and do not enhance after intravenous contrast administration, whereas the marrow cavity of ossified cartilage behaves like fatty marrow elsewhere in the body. On DWI, normal cartilages have a low signal on b1000 images and on ADC maps. These characteristic features on MR imaging and the increased capability of MR imaging to distinguish between tumor and peritumoral inflammatory changes greatly facilitate the assessment of laryngeal cartilage abnormalities and thus avoids diagnostic pitfalls related to CT.

MR Imaging Findings in Laryngeal and Hypopharyngeal Cancer

Tumor and Peritumoral Inflammation

Both primary and recurrent laryngeal and hypopharyngeal SCCs show characteristic signal intensity patterns on MR imaging: an intermediate signal intensity on T1, T2, and STIR and moderate contrast enhancement after intravenous administration of gadolinium chelates. Primary laryngeal and hypopharyngeal SCCs can be mass-like, they can extend along the mucosal surfaces or they can be diffusely infiltrating at cross-sectional imaging, whereas recurrent SCCs tend to be located beneath an intact mucosa and they display a diffusely infiltrative invasion pattern with indistinct tumor margins. This difference can be explained by histologic characteristics as primary laryngeal and hypopharyngeal SCCs have a rather unicentric growth pattern while recurrent tumors have multicentric tumor foci disseminated over large anatomic areas beneath an intact mucosa. Restricted diffusion is characteristic for both primary and recurrent SCCs and mean ADC values (calculated with b 0 and b 1000) are in the range of 0.9 to 1.3 10-3 mm2/s. Correlation with histopathology has shown that mean ADC values in laryngeal and hypopharyngeal SCCs are significantly correlated with cellularity, stromal component, and nuclear-cytoplasmic ratio, and differences in ADC values between tumors can be mainly explained by variable amounts of tumor stroma components.

Both primary and recurrent SCCs are often surrounded by variable degrees of peritumoral inflammation, which can accompany early and advanced cancers and may lead to overestimation of tumor size unless careful analysis of multiparametric MR imaging signal intensity is done. These peritumoral inflammatory changes which precede tumor invasion of adjacent structures are characterized by a higher signal intensity on T2 and STIR and by a stronger contrast enhancement than the tumor itself. Also, on DWI peritumoral inflammation shows no restricted diffusion, and mean ADC values (calculated with b 0 and b 1000) are typically in the range of 1.4 to 1.9 10-3 mm2/s.

Using a multiparametric approach with the aforementioned diagnostic criteria, MR imaging with current state-of-the-art coils allows an improved distinction between tumor and peritumoral inflammatory changes, therefore facilitating a more precise evaluation of critical anatomic subsites. Although newer studies found a low CT sensitivity for the detection of neoplastic invasion of the paraglottic space, older studies reported a high sensitivity of MR imaging and CT for the detection of neoplastic invasion of the pre-epiglottic space and paraglottic space. However, as older studies have used different morphologic criteria that do not allow distinction between tumor and inflammation, the reported specificity of MR imaging and CT in the paraglottic space was low, thus leading to overestimation of tumor spread in a considerable number of cases. Nevertheless, using the aforementioned MR imaging criteria, which take differences in signal intensity, contrast enhancement, and DWI characteristics into consideration, it is currently possible to distinguish between tumor and inflammation in the paraglottic space and in the TCAS with a sensitivity of 100% and a specificity of 78%. This area, which has important prognostic implications, is particularly difficult to evaluate on CT scans. Although other areas of the larynx, which contain higher amounts of fat, such as the pre-epiglottic space, are more easily evaluated with CT, it is worthwhile mentioning that currently there are no series in the literature systematically comparing the diagnostic performance of multiparametric DWI MR imaging with CT in the deep laryngeal spaces. Furthermore, because of the difficulty to obtain radiologic-pathologic correlation, most data in the literature are based on older studies, which have used older diagnostic criteria, and newer studies using the aforementioned criteria are based only on a limited number of cases.

Regarding the detection and staging of early SCC of the larynx and hypopharynx, data in the literature are scarce and do not allow to draw solid conclusions about the added value of MR imaging in T1 and T2 lesions. Nevertheless, some authors have pointed out that MR imaging is clearly superior to CT in the preoperative staging of early glottic cancer. It has even been suggested that DWI may detect changes in tumor size and shape before they become apparent at laryngostroboscopy and may help to distinguish laryngeal SCC from precursor lesions.

Neoplastic Cartilage Invasion

Detection of neoplastic cartilage invasion at CT uses the criteria of sclerosis, erosion/lysis, and extralaryngeal spread. Sclerosis is a sensitive but nonspecific sign of cartilage invasion, especially in the thyroid cartilage. In contrast, erosion/lysis and extralaryngeal spread are specific signs but not sensitive as they are bound to the presence of more advanced invasion of laryngeal cartilages. In the presence of sclerosis alone, it is nearly impossible to decide, whether sclerosis corresponds to cartilage invasion, whether it is caused by peritumoral inflammation with bone remodeling, or whether it corresponds to asymmetric ossification. Furthermore, it has been shown that the positive predictive value of CT to detect major cartilage invasion and extralaryngeal spread is limited and varies between 53% and 81%. Another problem with CT is that nonossified cartilage can have similar attenuation values as SCC on contrast-enhanced CT rendering a correct radiologic interpretation difficult. Applying the diagnostic criteria for tumor, inflammation, normal non-ossified, and normal ossified cartilage as explained earlier, MR imaging allows an improved diagnosis of cartilage abnormalities thus avoiding diagnostic pitfalls related to CT. In summary, if a cartilage in the immediate tumor vicinity displays a moderately high signal intensity on T2, a moderate enhancement on contrast-enhanced T1 and restricted diffusivity, the cartilage should be regarded as invaded. However, if a cartilage in the immediate tumor vicinity displays a higher signal intensity on T2 and a stronger enhancement than the adjacent tumor, and if there is no restricted diffusion, the diagnosis of peritumoral cartilage inflammation should be made. It is worthwhile mentioning that DWI images are not always contributive to the diagnosis due to geometric distortion and lower spatial resolution. Therefore, if DWI images are of insufficient diagnostic quality, in the authors’ experience, one should rely on standard morphologic MR imaging images.

Differentiating Tumor Recurrence from Post-treatment Changes

Recurrence in laryngeal SCC is relatively common and depends on age, subsite, stage, histologic differentiation, and treatment modality. Although the recurrence rate is about 5% to 13% in T1 cancer, T3-T4 cancers have a recurrence rate of about 30% to 40%. Recurrence most often occurs at the site of the primary tumor and about 90% of recurrences occur within 3 years after primary treatment. In hypopharyngeal SCC, recurrence rates are even higher than in laryngeal SCC and they equally depend on subsite, stage, histologic differentiation, and primary treatment modality. Early detection of recurrent disease plays a major role in a successful disease outcome. Endoscopic and clinical follow-up may overlook recurrent disease, especially after radiotherapy because of radiation-induced edema, fibrosis, or radiation-induced complications, for example, soft tissue necrosis or cartilage necrosis. Biopsy itself can substantially aggravate the situation by precipitating complications due to poor wound healing after biopsy. Furthermore, as recurrences after radiotherapy tend to occur under an intact mucosa, endoscopic biopsy may also miss recurrent disease because of the necessity to obtain deep biopsies without being able to identify endoscopically the most appropriate site for tissue sampling.

Several authors have demonstrated the added value of DWI for the detection of residual/recurrent head and neck cancer after radiotherapy in comparison to CT and for distinguishing residual/recurrent disease from benign post-treatment changes; however, they have pointed out that false-positive evaluations caused by late fibrosis/scar tissue still occurred. Other authors found that major overlap of ADC values measured in benign post-treatment changes and in recurrent laryngeal cancer limited the ability of quantitative DWI to distinguish between the two entities. Nevertheless, by carefully combining morphologic MR imaging criteria with DWI, a high diagnostic performance for distinguishing post-treatment residual/recurrent disease from benign changes after radiotherapy (in particular late fibrosis) can be achieved, the positive predictive value and the negative predictive value of DWI MR imaging being as high as 92% and 95%, respectively. Both late fibrosis/mature scar and residual/recurrent disease have low ADC values. This diagnostic DWI pitfall can be avoided as their morphologic characteristics are different: recurrent SCC has a moderately high signal intensity on T2; however, late fibrosis/mature scar typically displays a very low signal intensity on T2 (lower than or similar to the signal intensity of muscles) and often no or only minor contrast enhancement.

Differential Diagnosis

Less than 5% of laryngeal and hypopharyngeal tumors are of nonsquamous cell origin. Unlike primary SCC, they are often located beneath an intact mucosa and sampling errors may occur with endoscopic biopsy. The role of imaging mainly consists in confirming a submucosal tumor, determining the precise tumor extent, and guiding the endoscopist to the most appropriate biopsy site to avoid false-negative biopsies. Although some tumors, such as chondrosarcoma, lipoma, schwannoma, melanoma, paraganglioma or vascular lesions (hemangioma and vascular malformations) show characteristic imaging features at MR imaging allowing distinction from SCC, other tumor types show overlapping features with SCC and only deep targeted biopsy can differentiate between SCC and non-SCC. Such tumors include adenoid cystic carcinoma, adenocarcinoma, rhabdomyosarcoma, and many more. When lymphoma first presents as a submucosal laryngeal lesion without associated adenopathy (rare presentation), the diagnosis can be quite challenging. However, although the very low ADC values may suggest the diagnosis of lymphoma, biopsy is always required as other rare conditions, such as Rosai Dorfman disease or IgG4 related disease may present with similar MR imaging features. Nevertheless, as a general rule, in patients with a suspected lesion beneath a completely intact mucosa at endoscopy and a tumor mass with similar imaging characteristics as SCC, the radiologist should suggest the diagnosis of an unusual histology. It is also worthwhile mentioning that some non-neoplastic conditions, such as tuberculosis, may mimic SCC clinically and at morphologic MR imaging. The diagnosis can be quite challenging in nonendemic areas and in the absence of a known history of tuberculosis or predisposing factors. However, absent restriction of diffusion and pulmonary involvement in these cases are very helpful in suggesting the diagnosis and contributing to an adequate patient work-up.

What the Referring Physician Needs to Know

The major goal of radiology reports is to provide accurate, timely, and pertinent information. In many institutions, structured reporting is increasingly used for improved comprehensiveness, to avoid diagnostic errors by omitting key information, for a more consistent evaluation at multidisciplinary tumor boards, for follow-up purposes, and for data collection and research. Structured reporting includes a structured format, consistent organization, and consistent terminology. Even if an IT-based structured reporting template is not available at a radiologist’s institution, the following key information using standard terminology as described earlier should be included in every MR imaging report dealing with laryngeal and hypopharyngeal cancer:

• Which anatomic subsites of the larynx/hypopharynx are involved (the AJCC/TNM nomenclature should be used)?
• Is the pre-epiglottic space invaded?
• Is the paraglottic space invaded? If yes, which sites precisely?
• Are laryngeal cartilages invaded? If yes, to what degree (inner cortex vs through outer cortex)?
• Is there invasion of other structures beyond the larynx/hypopharynx, for example, strap muscles, soft tissues of the neck, carotid arteries, prevertebral space, or esophagus?
• Are there unilateral/bilateral lymph node metastases? What levels are involved?

In addition, at the authors’ institution, the reporting radiologists also indicate how confident they are with respect to involvement of the key anatomic areas mentioned earlier, in particular if there are technical issues impairing a confident interpretation of imaging findings. Furthermore, when certain key findings are reported, we indicate the series and image numbers and we provide appropriately annotated key images. This procedure is especially useful for follow-up examinations and for coherent evaluation at multidisciplinary tumor boards. It is equally important to stress the fact that lack of appropriate clinical history related to previous treatment and lack of images from previous radiological examinations can lead to diagnostic uncertainty and, therefore, only a close cooperation with the referring physicians will ultimately lead to a high-quality radiologic report.

Summary

The advent of high-resolution surface coils combined with parallel imaging techniques, more robust DWI techniques, and refined diagnostic criteria have been some of the most significant advances in oncologic imaging of the larynx and hypopharynx. Although MR imaging of the larynx and hypopharynx presents some technical challenges, advantages over CT outweigh disadvantages. Current state-of-the-art MR imaging allows a more precise assessment of submucosal tumor spread by detecting subtle soft-tissue abnormalities in anatomic areas that are more difficult to interpret on CT scans, such as the paraglottic space, the laryngeal cartilages, and the extralaryngeal soft tissues. Current MR imaging diagnostic criteria combining distinct signal intensity patterns on morphologic sequences with DWI features allow an improved discrimination between tumor, peritumoral inflammation and scar tissue, improved assessment of laryngeal cartilage abnormalities, and, ultimately, an increased precision for tumor delineation beyond the capability of multislice CT. Improved tumor delineation allows tailored treatment options, such as deciding between highly focused radiotherapy, transoral laser microsurgery, open partial laryngectomy, or total laryngectomy. Furthermore, MR imaging has a higher diagnostic performance than CT for the detection and precise depiction of post-treatment recurrent disease.

This text was extracted from the original article.