The use of a battery of examination methods for detection of cervical metastases in squamous cell carcinoma of the oral cavity

Introduction. In patients with squamous cell carcinoma of the orofacial area, the presence of cervical metastases represents a single most significant prognostic factor. This fact underlines the importance of thorough examination of the cervical lymph nodes for potential tumor involvement. To verify this, the most common investigative methods are physical examination (PE), sonography (US) and computed tomography (CT), which have also been used to assess the stage of the disease in the patients in our research. Objective. To evaluate the performance of individual methods (physical examination, sonography, computed tomography) and combinations. Method. Patients with squamous cell carcinoma of the oral cavity, who had undergone physical, US and CT examinations at our department followed by radical neck dissection were included in this retrospective study. A total of 57 patients were included. Results. The sensitivity of PE, US and CT were 38%, 69% and 61%, respectively, however CT+US combination yielded 83% sensitivity and combination of all these methods 86% sensitivity. The number of false positives was however relatively high with specificity of the 3-way combination at 65%. Conclusion. A combination of our three widely available inexpensive methods detected 86% of metastases in cervical nodes. The large number of false positives however indicates that the method should rather be used for screening in selecting patients who need additional and more expensive imaging than for diagnosing cervical metastases. Also, as 14% of cervical metastases pass undetected using our method, we would recommend an additional examination at least by US+PE several weeks to a few months after the initial examination.


INTRODUCTION
Tumors of the oral cavity followed by tumors of the nasal cavity, paranasal cavities, salivary glands, larynx and pharynx represent a major group of relatively frequent malignancies -the head and neck tumors. This group occupies the 7th place worldwide in tumor-related mortality. According to the IARC (International Agency for Research on Cancer), the global incidence of the tumors of the oral cavity and lips was in 2012 over 300,000 cases, i.e, approx. 2.1% of all tumors. The 2012 mortality related to these tumors was just below 150,000. The Age Standardized Incidence Rate according to the world standards (ASRW) was 4 cases per 100,000 population. Both the incidence and mortality in men was approx. twice as high than for women 1 . Squamous cell carcinoma is the dominant type of carcinoma in the oral cavity because most risk factors affect the upper layer of oral mucosa cells. Alcohol and tobacco abuse (including chewing or snuffing tobacco) are the most significant risk factors 2,3 . At least 75% of the tumors of the head and neck are caused by tobacco and alcohol abuse [4][5][6][7] . The synergistic effect of these risk factors has also been reported as people abusing both tobacco and alcohol are in higher risk of developing a tumor than people abusing just one 8,9 . Additional risk factors for developing spinocellular carcinoma may be poor dental hygiene or missing teeth 10,11 . Using a mouthwash with a high alcohol content is a possible, though not proven, risk factor 10,11 . The presence or absence of metastasis in the cervical lymph nodes is an important predictor of the prognosis and survival of patients with orofacial tumors. A one-sided lymph node metastasis reduces the survival rate by about 50% and bilateral metastasis by another 25% (ref. [12][13][14][15]. A major risk is occult metastases, the occurrence of which is closely related to the methods used for the examination of the cervical lymph nodes, with higher sensitivity of the used imaging method reducing the risk of the occurrence of occult metastasis 16 . The most common methods of examination of the cervical metastases include physical examination (PE), ultrasonography (US), computed tomography (CT) or magnetic resonance imaging (MRI) or modern methods such as fine needle aspiration cytology (FNAC) or positron emission tomography (PET). While the first three methods are relatively cheap and widely available, MRI, despite being also widely available, is much more expensive and FNAC or PET, although they can provide better results, combine the disadvantages of the high costs and limited accessibility in everyday practice. According to the recommendations of the Royal Society of Radiology, only CT is recommended for the initial staging of the tumor in the oral cavity, however MRI or PET are also mentioned as additional methods. US is however not mentioned in these guidelines et al. 17 . In this retrospective study, we aimed to evaluate the diagnostic accuracy of the methods used in our hospital (i.e., physical examination, ultrasonography and computed tomography) and to investigate whether a combination of some or all of those relatively inexpensive methods improves the diagnostic accuracy of metastases in the cervical lymph nodes.

METHODS
Our study group comprised patients who were treated for spinocellular carcinoma (verified histopathologically) of the oral cavity between 2011 and 2018 at the Department of the Oral and Maxillofacial Surgery at the University Hospital Ostrava and who had undergone examinations by all three investigated methods (PE, US, CT) with subsequent block neck dissection (BND) (Fig. 2) and histological examination of the removed lymph nodes. The individual neck spaces I-V (ref. 12,13 ) were after the excision marked in a way allowing the pathologist to evaluate metastases in the individual parts. The results of the examination methods were subsequently compared to the results of the histopathological examination and statistically evaluated. The research group comprised altogether 57 patients, 25 females and 32 males, mean age 59 years (33-85 years). The standard physical examination included palpation and visual evaluation of the site, size, consistency, mobility and character of the indurated surface. Imaging methods distinguished between benign (oval shape, below 10 mm in the long axis, echogenic hilum and perinodal fat tissue) and suspected (enlarged over 10 mm, round shape, extranodal extension, or necrotic changes) lymph nodes 18,22 . Ultrasonography examinations were performed using Aloka ProSound 3500 (Ohio, USA) or GE Logiq S7 (Massachusetts, USA) instruments. Linear probes with frequencies 3.5-13 MHz were used in the mode of soft tissue detection; parameters of the examination were customized by the personnel according to current acoustic conditions in the tissues. CT examinations utilized multidetector scanners Siemens Definition AS+ or Definition AS (Munich, Germany). The scanning was performed in thin collimation with image reconstruction in the soft tissue filter. In each patient, the findings were evaluated both in the native image and after intravenous application of appropriate quantity of iodine-containing contrast medium. In our hospital, a dedicated team consisting of an experienced oncologist, maxillofacial surgeon, and radiologist specialized in the imaging techniques of the head and neck was formed for evaluation of the nodal metastases. A clinical classification (cTNM) of the tumor was performed by this multidisciplinary team after the physical examination and imaging methods. As the inclusion criteria imply, all patients in our group underwent BND with subsequent histopathological evaluation (pTNM). The results were evaluated based on the presence of malignancy in individual cervical spaces. Confusion tables for individual examination methods as well as their combinations were subsequently prepared using histopathological findings as true data and sensitivities, specificities, positive and negative predictive values were calculated. If there was a suspention for malignity based on at least one examination, the lymph node was considered as malignant-i.e., clinically negative nodes were only those without any suspicion in any method. Patients signed informed consent with participation in the study.

RESULTS
There were 57 patients included in the study. The most common site of the primary carcinoma in our study was the floor of the oral cavity (30%) closely followed by the tongue (28%). Most of the tumors in our group were clinically classified as cT2 (30 patients; 53%) which was in good accordance with the histopathological classification which determined 28 patients (49%) to be in the T2 stage. We also evaluated the occurrence of the affected lymph nodes in the individual neck spaces -Ia, Ib, II, III, IV, a V. The most frequently affected nodes were in the Ib and II spaces, see  Table 1 shows the results of comparison between individual methods of clinical examination (or their combinations) and results of the histopathological examination. Table 2 presents standard test characteristics used for the evaluation of the individual clinical examination methods and their combinations related to the results of histopathological examination used as true data.

DISCUSSION
The degree of metastatic spread reflects the tumor aggressiveness and constitutes an important prognostic factor in relation to survival 19 . Nevertheless, is it not easy to clinically detect metastatic nodal involvement, because they can be mistaken for reactive inflammatory changes 20 . Even small lymph nodes clinically indistinguishable from healthy ones can contain malignant cells (micrometastases) that are only detectable during histopathological examination 20 . From this perspective, it might appear reasonable to perform block neck dissection (BND) in all cases of the spinocellular carcinoma. On the other hand, any unnecessary BND increases the patient morbidity and risk of post-operative complications. Hence, an accurate examination method would be very valuable in identification of patients benefiting from BND (ref. 21 ).
Although modern imaging methods such as PET/CT or FNAC became available over the last decades 11,22 , physical examination along with US and CT, along with MRI, which is however significantly more expensive, remain the principal examination methods. In this study, we have investigated the accuracy of these three methods (PE, US, CT) alone as well as of their combinations for detection of malignant nodes. All are relatively cheap and available but their individual sensitivity is quite low. The sensitivity of the physical examination alone in our group was 38%, which means that this method would fail to capture most malignant nodes. From this point of view, we cannot recommend using this method alone. Comparing CT and US, the results were similar, capturing 69% and 62% of malignant nodes, respectively. However, the combination increased the sensitivity to 83% and a combination of all three methods to 86%. Hence, a combination of all three methods shows the best results, however 14% of nodal metastases remains undetected. The good news is the high negative predictive value of approx. 97% for the combination of both US+CT and US+CT+PE. The high number of false positives is attested by the low positive predictive value (below 30%) and specificity (below 70%). This result cautions us that the combination is not sufficient to indicate a patient to BND as it often does. From this perspective, it appears that we should not perceive these methods and their combination as a diagnostic method but rather a relatively cheap Tier I screening battery to select patients for subsequent more expensive examinations such as PET, MRI or FLIR. It might still be beneficial to repeat at least the two cheapest methods associated with no radiation risk 1-2 months after the original examination, even in patients who turned out negative result during the original examination to minimize the risk of micrometastases, which could be presented at the time of the original examination. A study by Shetty et al. 22 investigated, as did our study, the performance of palpation, ultrasonography and CT for detection of metastatic cervical lymph nodes. Their results of physical examinations were almost identical to ours (36.6%), however the results for US (54.5%) and CT (31.8%) were poorer than ours. Rottey et al. 23 25 and Sarrion-Perez 26 with varying results. Detailed review is not within the scope of this paper and we would like to refer this review to interested readers. We are not aware of any study investigating the performance of a battery of examination methods such as that presented here.

CONCLUSIONS
Our study confirmed that individual widely available and relatively cheap imaging methods provide similar sensitivities and specificities for detection of potential cervical metastases of spinocellular carcinoma. Although the contribution of physical examination to the performance of a US+CT combination is relatively small, we should bear in mind that it is still the basic examination method that should be performed in every patient, so its inclusion in the battery of tests is reasonable. The PE+US+CT combination yielded an 86% sensitivity and 97% NPV, however the number of false positive detections was quite high. From this point of view, such a combination appears to be, due to its relatively low price, a suitable screening method useful for preselecting patients in need of additional, more expensive, imaging methods. Still, with a sensitivity of 86%, we also recommend further examination of patients at least using physical examination and US as our battery does not lead to capture of all metastases at the time of the original examination. We also believe that our hospital multidisciplinary team, including a maxillofacial surgeon, radiologist and oncologist, who are specialized in head and neck cancer, is an advantage.