Abstract
Standard magnetic resonance imaging (MRI) and computed tomography continue to be the imaging modalities of choice in staging and reviewing patients with head and neck cancer. Diffusion-weighted MRI (DW-MRI) is an advanced imaging modality that records the molecular diffusion of protons and thus provides an opportunity to further assess tissue character. Interest in DW-MRI of the head and neck continues to grow, especially its application to the assessment and treatment of head and neck cancer. We highlight the potential role of DW-MRI in the delineation, characterization, and lymph node staging of head and neck tumours. Furthermore, we discuss the ability of DW-MRI to provide a real opportunity to differentiate post-treatment tumour recurrence from chemoradiotherapy-induced local tissue changes. The future impact of these findings upon the clinical practice of the head and neck surgeon is discussed.
Introduction
For the modern head and neck surgeon, the ability to produce radiological images both sensitive and specific to head and neck malignancy is an invaluable tool. In this area, it is accepted that standard computed tomography (CT) and magnetic resonance imaging (MRI) have a high diagnostic accuracy in detecting cancers of the head and neck . However, both standard CT and MRI are anatomical-based imaging modalities and although this solves most diagnostic questions, there are additional imaging techniques that aim to evaluate the metabolic activity of abnormal tissue. Advanced MRI techniques such as spectroscopy, perfusion imaging, and diffusion-weighted (DW) imaging have been evaluated for their ability to characterize tissue and differentiate malignancy.
Standard MRI produces an image by recording the signal (resonance) generated from water protons as they re-align and ‘relax’ back to their original position after the application of a directional electromagnetic field causes the protons to ‘spin’. This rate of relaxation is variable for different tissues. DW-MRI records the molecular diffusion of protons corresponding to Brownian motion within biological tissues. This is achieved by applying pairs of opposing electromagnetic field gradients to the protons along a certain diffusion direction. The protons are ‘dephased’ by the first gradient and ‘rephased’ by an equal opposing secondary gradient, with only stationary molecules being totally rephased. Therefore, the fast movement of protons results in incomplete rephasing and is translated into a signal intensity decrease on the resulting Image. The differing sensitizing effects of the gradient (termed ‘b-value’), together with the speed of proton movement, will affect the signal intensity generated. Combining this signal intensity measurement with differing b-values, the apparent diffusion co-efficient (ADC) is calculated. The ADC value is the parameter that is used to quantify DW-MRI and it estimates the amount and speed of proton movement within the tissue. Diffusivity is affected by cell size, density, and integrity. Hence, dense hypercellular tissues such as tumour masses have reduced diffusion and therefore show low ADC values, whereas tissue changes such as oedema, inflammation, or fibrosis have low cellularity and show high ADC values. In malignant lesions, the increased microstructural density due to increased cell number, pleomorphism, and cell volume, restricts diffusion and therefore decreases DW-MRI ADC values when compared to the ordered tissue architecture of benign tissue.
At low b -values, the signal intensity may be influenced by microvascularity and perfusion of tissue. Higher b -values reflect truer changes in tissue diffusivity, and imaging is generally performed with b -values of 800 or 1000 (s/mm 2 ) on a clinical basis. However, higher b -values may adversely affect the image quality (signal-to-noise ratio) 2 . Diffusion imaging may be performed on 1.5 T and 3 T MRI systems and is matched to the anatomical sequences, generally with 4-mm sections in the axial plane. DW-MRI may be either visually assessed on a qualitative basis for differences in signal intensity, or the mean ADC value may be calculated by placing ‘regions of interest’ around the lesion.
To date the clinical application of DW-MRI has been realized in neurology and the imaging of intracranial lesions, predominantly acute stroke. Recent advances in MRI technical performance have allowed a growth in potential applications of DW-imaging at extracranial sites, with interest in the head and neck region continuing to grow. In this review we discuss the use of DW-MRI within head and neck oncology, focusing on the diagnostic application of this imaging modality in regard to head and neck tumour detection, characterization, and metastatic lymph node staging. Further, we discuss the role of DW-MRI in predicting treatment response and tumour recurrence.
Tumour delineation and characterization
Head and neck squamous cell carcinoma (HNSCC)
Wang et al. were among the first to use DW-MRI to characterize soft tissue head and neck lesions. In a cohort of 81 patients, with 81 lesions, they were able to demonstrate a statistically significant difference in the ADC values of malignant lymphoma, carcinoma, benign tumour, and cystic lesions (each having a higher ADC value than the last), with an accuracy of 86% to differentiate malignant lymphoma from carcinoma (92% sensitivity, 83% specificity). They postulated this imaging modality as providing useful information prior to surgery or biopsy.
Several further studies have provided evidence for the statistically significant difference in DW-MRI ADC values between benign and malignant head and neck lesions ( Fig. 1 ). One study evaluated 33 lesions (17 benign, 16 malignant) and was able to define discrete mean ADC values for benign (majority schwannoma or meningioma) and malignant lesions (majority HNSCC). The same group demonstrated significantly lower ADC values for HNSCC when compared to neck musculature, tongue, and thyroid tissue, but not when compared to parotid or submandibular tissue, thus raising questions regarding the diagnostic application of DW-MRI in the salivary tissues.
Further research has shown an ability not only to differentiate benign from malignant, but also to provide an analysis of histological grade at the time of imaging. In a recent article, Yun et al. analyzed DW-MRI ADC values of 54 HNSCC lesions with known histological grade (either from biopsy or surgery): 34 well-differentiated, 10 moderately differentiated, and 10 poorly differentiated lesions. They also evaluated standard and high b -values and the effect this had upon the ADC values of the lesions. Of note, the ADC values of well-differentiated and poorly differentiated lesions were significantly different across all b -value ranges, but not when compared to moderately differentiated lesions. This translated to poorly differentiated lesions appearing as a brighter signal intensity (lower ADC values) on DW-MRI images, and vice versa. Several other studies have noted a correlation between histological grade of HNSCC and DW-MRI mean ADC value, but failed to demonstrate a statistically significant difference, noting some overlap in ADC values. Therefore, the actual diagnostic merit of DW-MRI to establish HNSCC grade remains unknown and this will likely serve as a predictive adjunct to conventional diagnostic workup.
Salivary gland tumours
The success of DW-MRI to differentiate benign from malignant HNSCC has not been shared with salivary gland tumours, namely parotid gland tumours. The diversity of benign and malignant tumour groups within this tissue poses several challenges.
To date, two large studies have assessed DW-MRI of parotid gland tumours. These studies evaluated parotid gland tumours in 136 and 75 patients, respectively, comparing DW-MRI with standard MRI for histologically confirmed tumours. In the study of 136 patients, using mean ADC values, pleomorphic adenomas (most common tumour in the study, n = 43) were distinguishable from all other tumours except myoepithelial adenomas. However, the study was unable to provide a threshold ADC value to differentiate Warthin’s tumour (second most common tumour in the study, n = 32) from other tumours. While there was a trend for benign tumours to have higher ADC values than malignant tumours, there was significant overlap between the tumour groups and no discernible ADC cut-off value. In the more recent study of 75 patients, the group echoed the above findings of higher ADC values in malignant when compared to benign tumours, but again noted significant overlap in ADC values. This study also demonstrated a significant difference in the ADC values of pleomorphic adenomas compared to all other tumour groups. Furthermore, by proposing a set ADC cut-off value, they argued for the ability to discriminate pleomorphic adenomas from all malignant tumours (sensitivity 82.1% and specificity 81.2%).
These results are encouraging, with the apparent ability of DW-MRI to differentiate pleomorphic adenoma from other salivary tumours, given its risk of malignant transformation. The future diagnostic application and benefit of DW-MRI within salivary gland imaging is likely to be realized when used in combination with data of morphological characteristics derived from other imaging modalities.
Thyroid nodules
Differentiation between benign and malignant thyroid nodules can be difficult, and as an advanced imaging modality, DW-MRI has the potential to aid diagnosis in equivocal cases and may prevent or prompt further invasive investigations. Several studies have noted a significant difference in mean ADC values between benign and malignant thyroid nodules. The largest of these studies examined 93 thyroid nodules in 75 patients together with 20 healthy control subjects. Mean ADC values of malignant nodules were found to be significantly lower than those of benign nodules. It remains to be seen if DW-MRI of the thyroid will be able to differentiate tumour histological subtype. However, with a significant prevalence of thyroid nodules within the general population, Thoeny et al. speculate that DW-MRI of the thyroid may be of benefit to determine further imaging requirements in patients with incidentally discovered thyroid nodules on standard MRI.
Sinonasal, nasopharynx, base of skull, and masticator space imaging
With the large variation of sinonasal and skull base tumours, further CT and MRI criteria than those currently used are needed to evaluate benign from malignant lesions. White et al. were among the first to use DW-MRI to attempt to differentiate 13 types of benign and malignant lesions of the above anatomical areas (24 lesions in total) ( Fig. 2 ). While there was a statistically significant difference in the ADC values of benign and malignant lesions, there was also considerable ADC value overlap. They stated that while there is a likelihood that a lesion would be malignant with a lower ADC value, there remain several limitations. One such limitation is the artefact created by the skull base, which has been noted to hinder DW-MRI analysis of such lesions. Attempts have been made to overcome this by using non-echo planar techniques for the readout of the diffusion data, which is less prone to susceptibility effects. Fong et al. performed a direct comparison of ADC values in nasopharyngeal carcinoma (NPC), HNSCC, and extra-nodal head and neck lymphoma. In a cohort of 76 patients (45 NPC, 26 SCC, and 5 lymphoma), they demonstrated significantly different mean ADC values among the three groups. HNSCC exhibited the highest ADC, followed by NPC and then lymphoma with the lowest ADC value. However, as above, there was wide variation in ADC values of NPC and significant overlap between the NPC and HNSCC groups. This group argued that as SCC is rare in the nasopharynx, the main differential to NPC is lymphoma. As such, the high sensitivity and specificity (93.3% and 100%, respectively) to differentiate NPC from lymphoma provides evidence for the clinical potential of DW-MRI in this anatomical area.
With regard to the diagnosis of infection, DW-MRI has been investigated as a tool to differentiate solid malignant lesions from masticator space infection and skull base osteomyelitis. Abdel Razek and Nada demonstrated a clear difference in the ADC values of numerous malignant lesions in the masticator space when compared to infection (odontogenic, necrotizing otitis externa, and invasive fungal infections), with a statistically significant difference between infection and SCC or sarcoma. They argued that as both groups of patients may present with pain, swelling, and trismus, DW-MRI as a non-invasive investigation was useful to identify the rare number of patients with a malignant lesion. In a similar study, Wang et al. were able to define statistically significant mean ADC values between benign and malignant lesions in the masticator space. However, in contrast to the above group, they reported no statistical significance in the difference in ADC values of malignant and inflammatory lesions. Further to this, Ozgen et al. used DW-MRI to differentiate skull base osteomyelitis from malignant lesions (nasopharyngeal carcinoma, lymphoma, and metastasis). Mean ADC values of skull base osteomyelitis were significantly higher than those for nasopharyngeal tumours and lymphoma, but not when compared to metastasis (argued as being due to the variation in primary tumour).
Lymph node staging
With the presence of nodal metastases representing a significant adverse prognostic factor for long-term survival, the assessment of lymph node metastases and subsequent staging of head and neck cancer remains a vitally important research topic within head and neck surgery. The combination of clinical palpation and imaging remains the standard in head and neck cancer nodal assessment. However, its accuracy to detect small nodal metastases is low. While standard MRI is recognized as a good diagnostic tool in the nodal staging of HNSCC, it is not without its pitfalls. Assessment with standard imaging relies on size-related, enhancement, necrosis-related, and morphological criteria, and the detection of sub-centimetre nodes is technically challenging. Newer diagnostic techniques, such as sentinel node biopsy, are accumulating positive evidence; nevertheless, elective neck dissection often remains the treatment of choice in the clinically negative neck. As an advanced imaging modality, DW-MRI has the potential to provide non-invasive head and neck nodal staging ( Fig. 3 ).
Numerous studies have identified the diagnostic benefit of DW-MRI in cervical node assessment and staging. In a comparison of MRI and DW images of 219 cervical nodes in 16 patients, de Bondt et al. identified DW-MRI ADC values as the strongest independent predictor for the presence of nodal metastases in HNSCC, with 92.3% sensitivity and 83.9% specificity. Histological examination post neck dissection was used to determine nodal spread, with 26 nodes demonstrating evidence of metastases. When the ADC value was added to standard MRI diagnostic criteria (size and morphology), there was a significant improvement in metastatic node identification. Of note, in this study the majority of metastatic nodes were ≤1 cm in size (22/26) and the majority of patients (13/16) had a ‘clinically node-negative’ neck. In a larger study, Vandecaveye et al. evaluated nodal metastases in 33 patients with confirmed HNSCC, with a total of 76 metastatic nodes; 58% (44/76) of these were sub-centimetre in size. The ADC value for metastatic nodes was significantly lower than that for benign nodes, and using a set ADC threshold they recorded 91% accuracy, 84% sensitivity, and 94% specificity. To put into a clinical context their findings, they reported that with the use of DW-MRI there was a correct change of nodal stage for 36% (12/33) of patients. In three patients the stage was downgraded to N0, in four patients a contralateral metastasis was detected, and in six patients the nodal stage was upgraded. However, in four patients the nodal staging was incorrectly changed using DW-MRI values.
While standard CT and MRI remain the primary imaging modalities in the initial loco-regional staging of HNSCC, the role of DW-MRI appears to be as a complementary imaging modality to detect smaller sub-centimetre nodal metastases. In turn this may influence radiotherapy and surgical treatment planning and guide the decision to perform, or extent of, surgical neck dissection in the clinically node-negative neck.
Predicting treatment response and tumour recurrence
If dense tumour masses hinder diffusion and thus have low ADC values, once treatment is initiated (chemoradiotherapy) the breakdown and decrease in cell number and density should present as an increase in ADC value on DW imaging. Thus, evidence is mounting for the predictive value of DW-MRI to determine treatment response of head and neck cancers. King et al. investigated early ADC changes at 2 weeks after the start of chemoradiotherapy in 37 patients with stage III/IV HNSCC. Of this group, at 2 years, local failure to respond to treatment occurred in 16 patients, with 21 patients showing local control of the tumour mass. Tumours that responded to treatment displayed a significantly higher percentage increase in ADC value at 2 weeks compared to those that failed treatment. They also noted a trend that tumours with local failure exhibited higher pre-treatment mean ADC values than those responding to treatment. In a similar study, Vandecaveye et al. investigated ADC values in 29 patients with HNSCC (ranging from stage I to stage IV) undergoing primary chemoradiotherapy. In this study DW-MRI was performed pre-treatment and at 3 weeks after two cycles of chemotherapy (3 weeks apart) and completion of a course of radiotherapy. At the 2-year follow-up, 15 patients displayed complete disease remission. Corroborating the results of King et al., they demonstrated that the percentage ADC change at 3 weeks post-treatment was significantly lower for lesions with tumour recurrence than for those with complete remission. Compared to standard MRI and CT, DW-MRI at 3 weeks had both a higher positive predictive value and negative predictive value for early treatment response.
If measured throughout treatment there appears to be evidence for ADC values as an imaging biomarker for treatment response. One study performed DW-MRI on 30 HNSCC patients before and at 2 and 4 weeks during treatment, comparing this to tumour volume on standard MRI images, with a 2-year follow-up period. In this study, greater ADC changes during the early treatment phase correlated significantly with loco-regional control. This was not seen with standard MRI images. They demonstrated an accuracy of 95% if using a cut-off value of a 25% ADC increase at 4 weeks. In a similar study, King et al. corroborated these results of changes in serial ADC values determining treatment response. They identified that a fall in ADC during treatment identified patients who developed treatment failure with 90% accuracy, postulating that re-population of the mass with tumour cells causes a fall in ADC and may indicate a time point of tumour resistance to treatment.
Differentiating the benign post-treatment effects of chemoradiotherapy, local tissue oedema and inflammation, from residual or recurrent tumour poses a difficulty with standard MRI. Therefore, several studies have also sought to use DW-MRI to evaluate HNSCC patients post treatment. One study investigated suspected HNSCC tumour recurrence in a group of 26 patients at a post chemoradiotherapy mean time of 8 months, with DW-MRI compared to histopathological findings. This study reported 94.6% sensitivity, 95.9% specificity, and 95.5% accuracy when using ADC values to differentiate recurrent HNSCC from non-tumoural tissue changes. Recent work has shown promise using high b -value and multiple b -value (bi-exponential fitting) DW-MRI in detecting recurrent or residual tumour compared to post-treatment changes.
The value of DW-MRI as an early predictor of outcome and recurrence in HNSCC should be of interest to all members of the head and neck multi-disciplinary team (MDT). From a surgical viewpoint, reliable data on the early treatment response may assist in the selection of patients amenable to salvage surgery or to guide the timing and planning of definitive surgery in patients undergoing neo-adjuvant chemoradiotherapy.
Limitations and developments in diffusion-weighted MRI
While this article has detailed the potential clinical applications and advantages of using DW-MRI in the head and neck, it is not without its limitations. Firstly, while research seeks to define ADC ranges and thresholds for tumour characterization, the ADC thresholds defined in the literature cannot necessarily be extrapolated to other centres due to differences in magnet strength and imaging protocols. As such, each centre will need to obtain site-specific ADC cut-off values if this approach is to be applied on a routine clinical basis, and standardized technique and imaging protocols are therefore needed. Furthermore, the head and neck is a difficult area to image. The complex anatomy and presence of surgical or dental implants increases the risk of susceptibility artefacts, and there is a risk of motion artefacts due to jaw or neck movements. Technical improvements such as multishot echo planar imaging, non-echo planar imaging, stronger magnetic field gradient coils, phased array receiver coils, and parallel imaging have sought to minimize the above limitations.
A further limitation of current methods of assessing DW-MRI ADC values is that the analytical methods do not take into account the spatial information intrinsic within an image. There is early imaging evidence that further post-processing of images by using texture analysis may provide additional heterogeneity information that has prognostic and predictive significance. In head and neck cancers, texture analysis has been applied to improve tumour delineation for radiotherapy planning. It is also possible to calculate ADC with different b -values. For example, DW-MRI obtained using multiple low diffusion weightings ( b -values <150 s/mm 2 ) reflects microcapillary perfusion, which can be quantified by applying the principles of intravoxel incoherent motion, whereas higher b -values reflect true molecular diffusion. This is termed bi-exponential fitting, as opposed to the mono-exponential ADC analysis largely described in this article. By optimizing DW-MRI measurements using a range of low and high b -values, the tumour ADC may be calculated over the entire range of b -values to reflect tissue perfusion and diffusivity. Early results show that such an approach improves accuracy to both define tissue character and predict treatment response. It remains to be seen how far DW-MRI analysis can be advanced to improve tissue characterization in the head and neck.
In conclusion, this article has discussed how DW-MRI is able to reliably discriminate benign from malignant lesions in the head and neck, with further evidence of its ability to predict tumour type and research for its ability to stage tumours on-going. Arguably the greatest potential and immediate impact of DW-MRI upon the practice of the modern day head and neck surgeon is its ability to identify metastatic lymph nodes and to predict the response to chemoradiotherapy treatment. Such early prognostic information would guide both elective and salvage surgery. When compared to other advanced imaging modalities, such as positron emission tomography (PET)-CT, DW-MRI is cost-effective, non-ionizing, and can be added easily to a standard MRI protocol.
Further research is needed before DW-MRI can be used routinely and reliably in all head and neck centres. If adopted, DW-MRI should be used judiciously in combination with accepted imaging modalities, and a close MDT collaboration between radiologist, physicist, and clinician is essential. It is advised that ADC thresholds are established for each centre and differing MRI systems. Despite advances in research and surgical techniques, survival rates of head and neck cancer have not improved greatly over the past decade. Therefore, any non-invasive diagnostic tool that enables the surgeon to more accurately stage and to predict and monitor the treatment response of the patient with head and neck cancer should be welcomed and utilized.
Stay updated, free dental videos. Join our Telegram channel
VIDEdental - Online dental courses
