Imaging Tests for Predicting the Presence of Difficult Airway in Head and Neck Cancer Patients Undergoing Otorhinolaryngological Surgery

Patients with head and neck cancers represent a challenge for the surgical team from many points of view, but, especially, the surgical moment where greater stress generated corresponds to the perioperative management of the airway, because in many occasions we can face unexpected situations, most of the time, incidental findings can hinder ventilation and endotracheal intubation. Gutierrez et al., in 2018, decided to study four tomography measures and their correlation in anesthesia records with airway management difficulties. Material and methods: A retrospective, observational study was carried out in 104 patients operated by head and neck cancers over a period of 36 months, only in those with access to tomographic records. Four tomographic measurements were considered and were statistically related to the extreme degrees of visualization of the glottis (Cormack III–IV) and the presence of the physical examination of Mallampati III–IV. Results: After performing a multivariate model in the group of extreme degrees of visualization of the glottis, the results were not statistically significant (p > 0.05; 95% CI: 0.030–2.31: EPI/ PPW, 0.018–1.37 TB/PPW). In the Mallampati III–IV group, in the multivariate model only the VC/PPW showed clinically significant results (p < 0.05; 95% CI: 0.104–8.53). Conclusions: Tomographic measurements and the physical examination predictors could represent a useful guide in the prediction of the difficult airway in these patients.

Imaging Tests for Predicting the Presence of Difficult Airway in Head and Neck Cancer Patients… DOI: http://dx.doi.org /10.5772/intechopen.81606 lung cancer. In the USA, approximately 2000 cases occur each year, but this number is increasing in populations with high-risk ethnic origins settling in North America. Nasopharyngeal carcinomas are frequently associated with a latent infection of epithelial tumor cells with EBV, the etiological agent of infectious mononucleosis. Nasopharyngeal carcinomas are also associated with environmental and genetic factors in susceptible populations that have migrated to North America and still have an elevated risk of this disease. Unlike other epidermoid carcinomas of the head and neck, nasopharyngeal carcinomas can appear at an early age, with a maximum incidence evident in adolescents and young adults. Nasopharyngeal carcinomas are divided into three histological types according to the World Health Organization (WHO): the undifferentiated (WHO III) and the non-keratinized (WHO II) have latent infection by EBV in 95% of cases and are the most of the cases in North America and the rest of the world; the well-differentiated form (WHO I) is less frequent and accounts for 5% of cases worldwide, although in North America it represents 15-25% of cancers and is usually associated with traditional risk factors, such as smoking. Nasopharyngeal carcinomas are associated with a high risk of early regional lymph node metastases, prolonged natural evolution, and very high risk of distant dissemination [3].

Prevalence
In some cases, friable vocal cord tumors, subglottic stenosis, or tongue-based pathologies represent challenges for even the most experienced anesthesiologist. The incidence of difficult intubation in OC population is 7.1%. The American Society of Anesthesiologists defines "Difficult intubation" as the completion of multiple attempts of intubation in the presence or absence of tracheal disease [4].
We know, by difficult airway according to ASA, the clinical situation in which a conventionally trained anesthesiologist has difficulty in ventilating with facial mask, difficulty in tracheal intubation, or both [5]. The intubation failure plus ventilation failure occurs at 0.003%, being the most dramatic situation that can occur [6]. Intubation failure is the inability to place an endotracheal tube. Its incidence is 0.05% in the general population. The estimated incidence of all types of difficulties related to intubation or airway safety is estimated at 1-3% [7].
Currently, there are no generic references for the incidence of difficult intubation, due to the great diversity of material available in terms of video-laryngoscopy, which is nowadays commonly used. The incidence of failed intubation is around 0.05-0.35% [7].

Predictors of difficult airway
The preoperative recognition in physical examination can help to predict lifethreatening situations for our patients. The test of Mallampati is considered the gold standard of the predictors, relating to the structures of the base of the tongue with respect to the visibility of the after pharyngeal structures, on a scale from I to IV, in which III-IV associated with other predictors may suggest difficulty [6]. Lee et al. describe that the modified Mallampati test has a sensitivity to predict difficult intubation of 76% and a specificity of 77% [8]. On the other hand, it is suggested that the Mallampati test in the supine position is a better predictor than in the sitting position [9,10].
Numerous predictive tests of difficult airway can be mentioned with their different sensitivities and specificities that can improve the predictive power of the Mallampati: hyo-and thyromental distance [sensitivity (S) 88%/65% and specificity (E) 60%/81%, respectively], the bite test (S 88% and E 88%), circumference of the neck in the obese population (S 88.2% and E 83%), and mouth opening (S 26-47% and E 94%). Due to the statistical limitations observed in the studied populations, predictive models that seek to improve the statistical power of these physical examination findings have been developed from multivariate analysis. Mainly they can be mentioned: Wilson (S 75% and E 88%), El-Ganzouri (S 65% and E 94%), Arne (S 94% and E 96%), Karkouti (S 86% and E 96%), and Naguib original (S 81% and E 72%) [8]. Recall that Mallampati plus thyromental distance and sternum distance is the combination of physical tests with greater discriminative power, with a sensitivity of 100% and specificity of 92.7% [11].

Ultrasonography
The physical principles of ultrasound consist in the transformation of mechanical energy into electrical and vice versa. The ultrasound wave travels through the tissues undergoing phenomena of scattering, refraction, and reflection, with wave frequencies that oscillate from 2 to 20 MHz and therefore impossible to be heard by the human ear. The organs of the body have different amounts of water that can help us distinguish the images in the ultrasound monitor. This physical alteration of the wave is known as impedance and is usually measured as the resistance to propagation of sound waves from one medium to another. The greater difference in impedance between two tissues, the easier it will be to distinguish one from the other. The same happens with the air-water interface, and to improve it, we usually place gel in the ultrasound probe to reduce it and optimize the visualization of the tissues. That is why we usually see black (hypoechoic) air in the monitor, solid viscera with high water content in white (hyperechoic), and structures of intermediate echogenicity (isoechoic) [12].
Everything explained previously was not possible without the piezo crystals of the probes. These devices are capable of emitting an electric pulse on their crystals, which is transduced to mechanical energy, generating an "oscillatory" activity in the tissues, and the signal is again received by the probe and transformed into electrical energy giving an image on the monitor. This information is processed in terms of amplitude and time of return of the signal in the ultrasound equipment. Tissues are images interpreted with hypo-(black), iso-, or hyperechoic (white) characteristics in the ultrasound monitor [12].
The resolution of ultrasound devices can be divided into axial, lateral, spatial, and temporal. The axial resolution refers to the ability of the ultrasound probe to define two images. At higher wave frequencies, we will be able to appreciate more superficial structures with greater definition, but not organs that are deeper. The lateral resolution refers to the ability to differentiate two structures perpendicular to the ultrasound probe and depends on the position of the probe with respect to the structure to be evaluated [12].
Spatial resolution is a fixed property of the probe and refers to its ability to solve objects located at the same height or thickness of the ultrasound beam. The number of individual PZT crystals that emit and receive the ultrasound waves and their sensitivity affect the resolution, accuracy, and clarity of the image. The temporal resolution refers to the clarity or resolution of structures in movement [12].
Ultrasounds seem to be beginning to play a role in the perioperative prediction of intubation difficulties. Chou et al. studied hyomandibular distance measured by X-rays and said that it may be increased in patients with difficult airway. The cutoff values described in this study were 33.8 mm in men and 26.4 mm in women. Knowing the limitations of this work, in 2008, it was described that the inability to see the hyoid bone in ultrasound (due to a hypopharyngeal of the base of the tongue or by a short mandibular branch) seems to increase the statistical power in the usefulness of ultrasound in the patient difficult to intubate [13].
Based on the previous study, years later Hui et al. recruited 110 patients that were trained in sublingual ultrasounds after a series of simple instructions. The probability of visualizing the hyoid bone by ultrasound was related in patients with Cormack I and II with a positive probability ratio of 21.6, which suggests that it can become a useful tool to predict difficult airway [14] Figure 2.
However, the shape and size of the tongue added to the fact that the patients were not in the sniffing position during the measurements may represent some limitations in the interpretation of the results in this study [14].

Computed tomography
The image in computerized axial tomography is an axial and coronal representation of the patient using the physical principles of X-rays. The final reconstruction of the image will generate a series of overlapping structures with attenuations that will depend on the amount of water contained in the tissue under study [15].
The CT image is produced by the process of reconstruction, digitally combining information from X-ray projections through the patient from many different angles to produce the cross-sectional image. Because the image is digital, it is made up of a group of pixels (shortened from "picture elements"). Each pixel has a gray scale value that is displayed to the viewer. The image is 2D, but it represents a 3D volume of tissue with a finite thickness (usually a very small thickness compared to the fieldof-view (FOV) size [≈2-5 mm]). Each pixel is the projection, or 2D representation, of the X-ray attenuation of a voxel (shortened from "volume element") of physical tissue. The size of the pixels and the thickness of the voxels relate to some important image quality features, such as detail, noise, contrast, accuracy of the attenuation measurement (CT number value), and artifacts. These will be discussed in more detail as they relate to the processes of acquiring and reconstructing CT data [15].
In a CT of a single section of tissue using a single detector, the X-ray beam is collimated to the desired image thickness. The detector array has a number of individual detector elements that each records the intensity of the beam passing through the tissue along the path from the X-ray tube to the element. The system captures a simple projection X-ray through the patient, consisting of a thin strip or row of pixels. It can be thought of as a one-dimensional (1D) radiograph. The scanner then rotates the source and detector to capture additional 1D "strip X-rays" through the same section of the patient, viewed from a number of angles. Each strip radiograph (projection) is stored in the computer memory for later reconstruction [15].
In multislice CT, this operation is performed simultaneously for many arrays of detectors stacked side by side along the z-axis (long axis) of the patient. The X-ray beam collimators can be opened so that a wider section of the patient is irradiated, and each row of detectors can measure a separate transmission signal for the tissue section that lies between the detector row and the tube. The width of tissue that is sampled by each detector row is determined by the physical width of the detector elements along the z-axis [15].
According to X-ray absorption, we use the scale of the Hounsfield units (HU) that allows to stratify the pixels of the measurements according to their densities with respect to water. The advantage of the HU scale is that density differences of 1 part in 1000 (0.1%) can be represented by distinct values. The inherent density resolution of CT scanners is about 0.5%, so the HU scale is sufficient to display all attenuation differences the scanner can measure. Increasing the value of K would not improve on the density resolution of the system [15].
Many authors have tried to apply these imaging tests for the study of airway anatomy. Randell    considering by means of NMR and CT studies parameters such as length of epiglottis, length of tongue, and width thereof, as difficulty markers of the passage of the endotracheal tube during intubation with a fiberoptic bronchoscope [16].

Computed magnetic resonance
Understanding the physics of magnetic resonance is a challenge for many medical specialists, as it involves implicit technical knowledge shared with other professional fields; it may even be harder to understand than the basic principles of tomography and ultrasound. The four main points of the process are:

Preparation
The human being is basically composed of water. Hydrogen molecules contained in water when exposed to a "magnet" can be magnetized. Basically the resonator is a magnet, which confers spin to the hydrogen with a resulting vector that oscillates at a frequency called Larmor that is proportional to the magnetic field surrounding it. This vector is going to align with the said field when it is exposed to 1.5-3.0 T. Magnetization can be manipulated with the use of contrasts such as gadolinium through a process called inversion of the image [17].

Excitation
The "spinning" of the hydrogen molecules generates a radiofrequency pulse in the Larmor spectrum, which is received by a coil (electrical conductor) that is transverse to the tissue magnetization under study and transduced by Faraday thermal induction; the previous really will generate the "magnetic resonance signal." The signal is attenuated by two processes called relaxation. This usually occurs in two times (T). T2 represents the loss of coherence of the spin of the hydrogen molecules over time. T1 represents the time it takes for the vector of the magnetic field to reach equilibrium. The important thing about these last two concepts is that they will determine the resolution of the soft parts during the study [17].

Spatial encoding
The frequency of the spin of the hydrogen molecules with different spatial location differs from each other. The sum of these frequencies generates a resulting vector causing a force called Lorentz force, which is perceived by the coils and will participate in the genesis of the acoustic signal generated during a resonance.

Signal acquisition
The different spatial location of the hydrogen ions and their individual rotation frequency are analyzed by a Fourier transform, which will generate the different pixels of the image and is stored in the magnetic resonance equipment [17].
The application of these concepts to the perioperative management of the airway can be observed. In the work on predictors of difficulty of endotracheal intubation considered from NMR studies conducted by Samra et al. in 40 patients in 1995, no significant differences were found in the 20/21 parameters studied in soft tissue X-ray imaging. NMR is among the difficult intubation groups not anticipated and achieved. It seems that no radiological study alone is able to predict the difficulty in intubation; one of the reasons is that the exposure of the larynx depends on its compressibility. The tongue and the soft tissue consistency of the floor of the mouth, a situation that cannot be measured with imaging studies, perhaps dynamic studies, could be useful in these cases [18].

Imaging modalities: weaknesses and strengths
In head and neck cancers, tomography through the use of detectors (64 midrange) can provide dynamic images to assess phonation or oral cavity tumors. In addition, it is a useful rapid test in claustrophobic patients with respiratory problems [19,20].
Some authors prefer the use of NMR in their institutions, since it is superior in resolution to delimit tumors and bone/cartilaginous parts and perineural invasion and to reach areas such as the base of the tongue that can be difficult to see in nasofibroscopy [21,22] Figure 3.
The 4-7% of head and neck tumors has lung metastases. Many hospital protocols use CT with low levels of radiation to follow these patients and even to determine responses to treatment, tumor biology, and nodal extension. And we cannot forget how important it is in radiotherapy simulation by tomographic tests. So in this group of patients, it seems that the two tests continue to complement each other [23][24][25][26][27].
Currently, it seems that MRI greatly improves the visualization of the bone and cartilage. The importance of this fact is related to staging and treatment of these patients [28][29][30] but not so in patients on chemotherapy treatment where positron emission tomography remains a better alternative for follow-up at 3 months [31].
CT protocols are less varied than MRI, but the technique still requires optimization and patient cooperation. Many of these patients need a study that ranges from the base of the skull to the chest with a 1-1.5 mm collimation. The use of contrast offers an important advantage in these patients, and, whenever administered at 1 ml/s, it helps to define better vascular structures and squamous cell tumors. The patient is then asked to raise their arms above their head, and the entire chest is scanned. Dental artifact remains a challenge in a patient group who will often have poor dentition. An additional limited CT with angled gantry can be used to mitigate beam hardening if the tumor is obscured [32].
In summary, CT continues to be used in most head and neck cancer patients. The cost of NRM continues to be a limitation for many centers. The tomography as a basis for  PET and radiotherapy remains a very valuable option in these patients, since it facilitates staging and treatment. It seems that in the future, it is inevitable that the MRI replaces the CT because of its resolution of soft tissue visualization, and therefore it would offer more targeted treatments to this area of such a complex anatomy [33].

Any place for CT and airway?
Patient intervened for head and neck cancers presents anatomical particularities in the airway, which are probably better described in tests performed routinely in this population as the CT. Comparing the degrees of visualization of the glottis and Mallampati test described in the anesthesia records with four tomographic measurements, the authors of this chapter propose a tool to improve the perioperative management of the airway in this group of patients. A retrospective study was conducted with 104 patients operated for head and neck cancers under general anesthesia and endotracheal intubation in the Otorhinolaryngology department during a period of 36 months. Throughout the selection process, the radiology team reported a number of 15 cases with significant distortion of the airway that are being excluded from the analysis [1].
Based on the findings of the preoperative imaging tests, a multivariate logistic regression analysis was performed, where the dependent variable was the presence of Mallampati III-IV and the extreme degrees of visibility of the glottis (defined as Cormack III-IV). The 89 patients were assigned to the Cormack I-IV and Mallampati I-IV groups in the analysis in equal proportion. A total of four tomographic and clinical factors of difficult airway were introduced into this model. The tomographic predictors considered in the study were the following: distance from the vocal cords to the posterior pharyngeal wall (CVV/PFP) and laryngotracheal angle (Alaring) and distance from the epiglottis to the posterior pharyngeal wall (EPI/PFP) and from the base of the tongue to the posterior pharyngeal wall (BL/PFP). The odds ratio (OR) was 95% with confidence intervals (CI). All the tests were considered statistically significant for all data analyses when p < 0.05 [1]   The most frequent surgeries performed during the study period were laryngeal microsurgery, total laryngectomy, and laser cordectomy, with percentages of 37.1, 10.1, and 6.7%, respectively.
In the Mallampati III-IV group, the mean in millimeters of distances of CVV/ PFP and BL/PFP was 11.89 and 8.82, respectively. In the Cormack III-IV group, the means of EPI/PFP and BL/PFP were 7 and 10.40 mm. The tomographic predictors EPI/PFP and 177 BL/PFP in Cormack III-IV patients show significant results in the univariate model (p < 0.05, 95% CI: 0.125-3.84 and 0.654-5.915, respectively), but this situation was not repeated in the multivariate model when the variables were analyzed categorically (95% CI: 0.030-2.31 EPI/PFP, 0.018-1.37 BL/PFP). ROC curve was also assessed during the study, with 71 and 69% being observed for EPI/ PFP and BL/PFP, respectively [1] Table 1.
From the tomographic predictors, the CVV/PFP distance in the Mallampati III-IV patients shows significant results in the univariate model (p < 0.05, 95% CI: 0.032-3.682), a situation that was repeated in the multivariate model with the same distance (p < 0.05, 95% CI: 0.104-8.53) The diagnostic yield of CVV/PFP was also assessed during the study, with 64% being observed [1].
Although is important to consider that most of the intubations are performed by residents, there is a possibility that the interpretation of the Cormack degree could be altered. On the other hand, were described must take into account the preferences for airway devices according to the experience of each doctor, since this may condition the analysis of the variables. Discerning the differences between device preferences is beyond the scope of this study [1]. Patients undergoing oncological head and neck surgeries receive preoperative radiotherapy and other diagnostic-therapeutic techniques that were sometimes difficult to relate temporally with the CT evaluated. This fact may condition the interpretation of the imaging tests by the expert radiologist [1].

Conclusion
In certain patients, it seems that the role of ultrasonography in the prediction of difficult intubation plays an important role [34]. We cannot forget the importance of the predictors to physical examination, and that their sum complements their statistical power. The particularities of airway management in patients with head and neck cancers are well known, and sometimes supporting us in preoperative imaging tests, in conjunction with the interdisciplinary communication of the surgical team, favors the reduction of unwanted events in the management of the airway.
It is useful in these special populations to anticipate a possible difficult airway. Imaging tests, particularly CT, could help in this regard. Gutiérrez et al. described in the group Mallampati III-IV distance measurement CVV/PFP shows significant statistics in the multivariate model (p < 0.05, 95% CI: 0.032-3.682). However, obtaining images in a reproducible position that recreates the position that the patient will adopt during intubation and prospective evaluation in a larger population can provide more useful preprocedure information for the anesthetist [1].
Knowing the limitations of difficulty predictors of airway, it seems that techniques are necessary alternatives that support doctors and allow us to foresee life-threatening situations.