Improving Diagnostic Accuracy of Cervical Metastases With Computed Tomography and Magnetic Resonance Imaging Peter J.

Hillsamer, MD; David E. Schuller, MD; Robert B. McGhee, Jr, MD; Donald Chakeres, MD; Donn C. Young, PhD

\s=b\ Elective neck dissection in patients with head and neck cancer continues to be controversial. The management of these patients would be greatly facilitated by improvements in predicting cervical metastases. Recent investigations have suggested that computed tomography and

magnetic

resonance

imaging

are

more

sensitive in detecting cervical metastases than physical examination. The Department of Otolaryngology at the Ohio State University Hospitals, Columbus, under\x=req-\ took a prospective study to compare the preoperative sensitivities of physical examination, computed tomography, and magnetic resonance imaging with pathologic findings in 27 patients undergoing neck dissections for head and neck cancer. The results indicate that computed tomography and magnetic resonance imaging were more sensitive (84% and 92%, respectively) than physical examination (75%), although the results did not achieve statistical significance. The sensitivity of combined computed tomography and magnetic resonance imaging was 90%. (Arch Otolaryngol Head Neck Surg.

1990;116:1297-1301) Accepted for publication June 28, 1990. From the Departments of Otolaryngology (Drs Hillsamer, Schuller, and Young) and Radiology (Drs McGhee and Chakeres), The Ohio State University, Columbus. Presented at the American Society of Head and Neck Surgery, Palm Beach, Fla, April 30 to May 1, 1990.

Reprint requests to the Department of Otolaryngology, The Ohio State University, 4100 University Hospitals Clinic, 456 W 10th Ave, Columbus, OH 43210-1240 (Dr Schuller).

benefits, morbidity, The indications, mortality electively

and of treat¬ ing the clinically negative neck is an important issue.15 The "30 percent rule" of probability for occult disease has been arbitrarily and almost uni¬

versally accepted as a guideline to treat the clinically negative neck. It is

not the intent of this article to review the advantages and disadvantages of elective treatment, but rather to eval¬ uate the utility of imaging studies in the decision-making process for man¬ agement of the clinically negative neck. Recent investigations611 have sug¬ gested that computed tomography (CT) and magnetic resonance imaging (MRI) have a higher sensitivity for cervical metastasis than clinical ex¬ amination alone. To further clarify the added benefit of imaging, or which modality is best, a prospective study was undertaken in the Department of Otolaryngology at The Ohio State Uni¬ versity Hospitals, Columbus, from July 1, 1988 to November 1, 1989.

MATERIALS AND METHODS All patients with head and neck cancer who presented from July 1,1988 to Novem¬ ber 1,1989, and who were considered candi¬ dates to undergo therapeutic or elective neck dissection, were asked to participate in this prospective study. After informed con¬ sent was obtained, 42 patients were eligible

Downloaded From: http://archotol.jamanetwork.com/ by a Virginia Tech User on 05/14/2015

for the study. Of the 42 patients, 27 under¬ went CT and/or MRI (19 patients under¬ went both). Unavailability of patient ac¬ crual was secondary to scheduling difficul¬ ties of both imaging modalities in a preadmission fashion. Physical examination was performed by one of us (D.E.S.) and reported according to the TNM staging system of the American Joint Committee on Cancer (1988). Com¬ puted tomography was performed on three types of scanners (General Electric 9800 CT scanner, Technicare 1440 HPS CT scanner, and Technicare 2060 Quantum CT scanner). Five-millimeter-thick contiguous axial sec¬ tions were obtained during infusion of iodinated contrast. Criteria for cervical me¬ tastasis on CT included: (1) lymph node size greater than or equal to 15 mm in largest diameter, (2) nodes of any size with evi¬ dence of central necrosis, (3) grouping of three or more ill-defined lymph nodes mea¬ suring 8 to 15 mm, or (4) loss of tissue planes. Criteria for extranodal extension included: (1) ill-defined margins around ab¬ normal lymph nodes, (2) edema or thicken¬ ing of adjacent fat and muscle, or (3) loss of fasciai planes between a mass adjacent structure.12

Magnetic

resonance

imaging

was

per¬

formed using a receive only anterior sur¬ face coil on a 1.5-Tesla superconducting magnet (Signa, General Electric Medical

Systems, Milwaukee, Wis). T,-weighted sagittal images (TR 800, TE 20) and spin echo axial images (TR 1500-3000, TE 20-30, 60 to 80 milliseconds) were obtained using slice thickness of 5- and l-mm gaps. Coronal images using various pulse sea

quences were obtained in selected patients. Criteria for cervical métastases on MRI

the

those used for CT. Necro¬ sis in lymph nodes seems more difficult to detect on MRI than on CT, therefore this criterion is difficult to use with MRI. Imag¬ ing examinations were interpreted by four neuroradiologists at The Ohio State Uni¬ were

of Patient

Findings*

—

Patient

was performed by one of (D.E.S.). Twenty-two patients under¬

these values.

RESULTS

Table 1 outlines the findings of phys¬ ical examination, CT, MRI, and histo¬ logie examination on each patient. Twenty-seven patients underwent MRI and/or CT, and 19 of the 27 patients underwent both imaging mo¬ dalities. Eleven of 27 patients pre¬ sented with a clinically negative neck. Sixteen of 27 patients presented with palpable nodal disease. Eight of 27 pa¬ tients had recurrent regional nodal disease and had previously received radiation therapy. Primary sites in¬ clude parotid gland in two, oral cavity in six, oropharynx in eight, hypophar¬ ynx in two, larynx in three, and un¬ known primary in two. Computed tomography and MRI were found to be superior over physi¬ cal examination in sensitivity values,

false-positive values, false-negative values, and overall efficiency predic¬ tive values. However, the values did not achieve statistical significance us¬ ing the 2 test due to small sample size. Sensitivity values were 75% for phys¬ ical examination, 84% for CT, and 92%

for MRI (Table 2). The overall effi¬

ciency predictive values were 78% for physical examination, 81% for CT, and 90% for MRI. Of 11 patients who initially pre¬ sented with a clinically negative neck,

MRI

CT

Pathologic Findings

T2, NO T2, NO T4, N1 NO

T2, N3 T4, NO

Recurrent neck

went

examination, CT, or MRI. Statistical analysis was performed in¬ cluding sensitivity values, specificity val¬ ues, positive and negative predictive values, overall efficiency predictive values,10 falsenegative values, and false-positive values. The 2 test with Yates' correction was used to determine statistical significance of

PE

Stage

Oropharynx Recurrent oropharynx Oropharynx Oropharynx

Neck dissection

standard neck dissection. Four pa¬ tients had preservation of the spinal acces¬ sory nerve. One patient had a supraomohyoid dissection. Routine hematoxylin-eosin staining and histologie examination of the neck contents were used to verify the findings of physical

Site Parotid Oral cavity

versity Hospitals. us

Summary

Table 1.

same as

Unknown

N2b

primary

Recurrent neck

N2b

10

Oropharynx Unknown primary

T4, NO

12

Recurrent neck Recurrent or cavity Parotid

N3

14

15 16 17 18 19 20 22 23 24 25 26 27

NA

N1 N2b

T2, T4, T4, T2, T4,

Oropharynx Oropharynx Supraglottic Hypopharynx

NO N3b N2b N3a

NOa

Hypopharynx

T2, N3a T3, NO T3, NO

Larynx cavity

Recurrent neck Oral cavity Recurrent oropharynx

NA

NO

Recurrent neck

Oral

NA

T3, NO

NA

N1

Oral cavity Oral cavity

NA T4, N3a NA T4, NO *PE indicates physical examination; CT, computed tomography; MRI, magnetic resonance imaging; minus sign, negative; plus sign, positive; and NA, not available.

Table 2.—Statistical Data' No. PE

CT

(%) MRI

16/19 (84) Sensitivity 13/14 (92) 15/20(75) 6/7 (85) 5/7 (71) Specificity 5/6 (83) False positive 1/16 (6) 2/18 (11) 1/14 (7) False negative 5/11 (45) 7/8 (38) 1/6 (17) Positive predictive value 15/16 (93) 13/14 (92) 16/18(88) 6/11 (54) 5/8 (62) Negative predictive value 5/6 (83) 21/27 (78) Overall efficiency 18/20 (90) 21/26(81) PE indicates physical examination; CT, computed tomography; and MRI, magnetic

three (27%) were upstaged (positive nodal disease) with MRI or CT scan. There was an undetected preoperative occult rate of 5 (45%) of 11. The rate dropped to 2 (18% ) of 11 with preoper¬ ative CT scanning and MRI. Patient 2 presented with a T2, NO, MO stage II squamous cell cancer of the oral cavity. Computed tomography and MRI (Fig 1) demonstrated a submental métastases missed on physical examination. Patient 4 had recurrent oropharyn¬ geal cancer after radiation therapy.

Downloaded From: http://archotol.jamanetwork.com/ by a Virginia Tech User on 05/14/2015

CT and MRI 18/20 (90)

(100) (10) 4/10 (40) 18/20 (90) 6/10 (60) 24/30 (80) resonance imaging. 7/7

2/20

Physical examination of the neck re¬ vealed a woody indurated neck with no obvious cervical métastases but MRI and CT were suspicious for cervical métastases (Fig 2). Pathologic exami¬ nation confirmed cervical métastases. COMMENT The treatment of the clinically neg¬ ative neck in patients with head and neck cancer continues to be a challenge because of the difficulty in staging the neck. The physical examination alone has significant limitations. South-

Fig 1.—Clinically occult lymph node metastasis. Left, A contrast-enhanced axial computed tomographic scan at the level of the hyoid bone shows a 15-mm submental lymph node (white arrow). Central low attenuation suggests necrosis within the node. Portions of the normal submandibular glands can be seen (white arrowheads). Right, On the left, an axial 12-weighted spin echo magnetic resonance image (TR 2500, TE 80) at the same level as in Fig 1, left, demonstrates homogeneously high signal in the submental lymph node (black arrow). The presence of necrosis is not apparent. The normal submandibular glands are seen bilaterally (black arrowheads). On the right, a sagittal ,-weighted spin echo magnetic resonance image (TR 800, TE 20) just to the left of midline shows a homogeneous intermediate signal submental lymph node (white arrow) surrounded by high-signal subcutaneous fat.

métastases after radiation therapy. Left, A contrast-enhanced axial computed at the level of the floor of the mouth shows multiple masses with low attenuated centers (white arrows) ad¬ jacent to the left carotid artery (black open arrow) and internal jugular vein (curved arrow). Local recurrence at the floor of the mouth on the left (white arrowhead) obliterates the normal fat planes at the base of the tongue. There is encroachment

Fig 2.—Clinically occult lymph node

tomographic scan

the left oropharyngeal airway. Right, Proton density (left) and T2-weighted (right) axial spin echo magnetic resonance im¬ ages (TR 2500, TE 20/80) at the same level as in Fig 2, left, show a 2-cm mass (black arrow) representing lymphadenopathy adjacent to the left carotid artery (white open arrow) and internal jugular vein (white curved arrow). The nodes are of high signal on the T2-weighted image, and the presence of central necrosis within the nodes is not apparent on these magnetic resonance images. Normal flow void is seen in the left carotid artery. The left internal jugular vein is slightly compressed. Lo¬ cal recurrence at the floor of the mouth on the left (white arrowhead) is best seen as an area of high signal on the T2-weighted on

¡mage.

Downloaded From: http://archotol.jamanetwork.com/ by a Virginia Tech User on 05/14/2015

Fig 3.—Extracapsular carotid artery invasion. Left, A contrast-enhanced axial computed tomographic scan at the level of the thyroid cartilage shows a large mass (solid arrows) deep to and inseparable from the right sternocleidomastoid muscle. There is no loss of the fat planes around the right carotid artery and jugular vein, suggesting the possibility of carotid artery invasion. Round area of low attenuation within the mass (crossed straight arrow) could represent a thrombosed internal jugular vein. The normal left carotid artery (open arrow) and internal jugular vein (curved arrow) are surrounded by fat. Right, On the left, an axial T2-weighted spin echo magnetic resonance image (TR 2500, TE 80) at the same level as in Fig 3, left, demonstrates a homogeneous high-signal mass (straight arrows) within the right side of the neck, lateral to the strap muscles. There Is nor¬ mal flow void in the right carotid (arrowhead) indicating it is patent, although it is surrounded by the mass. The wall of the right carotid artery is poorly defined. There is an ovoid area of high signal (crossed arrow) within the mass that may represent a thrombosed internal jugular vein. The left carotid artery wall and central signal flow void (open arrow) and the left internal jug¬ ular vein (curved arrow) are normal. On the right, a sagittal T,-weighted spin echo magnetic resonance image (TR 800, TE 20) through the right side of the neck paralleling the carotid artery shows a large mass (straight arrows) surrounding a patent right carotid artery (arrowhead). The mass is so intense with the sternocleidomastoid muscle. There is nonvisualization of the internal jugular vein.

wick13 found a 30% false-positive rate and a 39% false-negative rate for phys¬ ical examination. Fisch14 reported on the impracticality of lymphangiogram in predicting cervical metastasis. Prior to the development of the newer imaging modalities, the clinician was forced to rely heavily on the results of the physical examination recognizing its insensitivity. The process for mak¬ ing therapeutic decisions based on this inaccuracy along with the associated morbidity and cost of treating the neck had to be accepted. The presence of nodal disease is an important prognos¬ tic factor in head and neck cancer.15 Proponents for electively treating the neck point to the advantage of in¬ creased survival and local regional control if the neck is electively treated before nodal métastases develop.316·17 Improving the predictability of cervi¬ cal nodal métastases would enhance treatment

specificity.

In 1981, Mancuso et al6 reported on the utility of CT in evaluation of cervi¬ cal cancer. His criteria for cancer were nodal diameter greater than 15 mm, indistinct or obliterated tissue planes

around the carotid sheath, grouping of three or more 8- to 15-mm nodes, cen¬ tral decreased density of a node, and poorly defined mass in a lymph node-

bearing area. Twenty-three patients had pathologic confirmation (13 of them were studied retrospectively and nine of them were studied prospectively). The CT scan was accurate in 21 (91%) of 23 patients. Initially, eleven were clinically negative and 6 (55% ) of 11 were upstaged to positive nodal dis¬ ease by CT. These were correctly up¬ staged by pathologic confirmation. No evaluation for statistical significance was performed in this study. Stevens et al8 presented 40 patients studied prospectively and found that CT was accurate (93% ) compared with physical examination (70%). Sixteen patients initially presented with a clinically negative neck (NO). Of these, five (31%) were correctly upstaged to positive nodal disease by CT. No anal¬ ysis of statistical significance was per¬ formed. Friedman et al7 reported on 50 pa¬ tients studied retrospectively and found physical examination to have an

Downloaded From: http://archotol.jamanetwork.com/ by a Virginia Tech User on 05/14/2015

sensitivity and CT to have a 90% sensitivity. No analysis for statistical significance was performed. Eighteen of the 50 patients presented initially with a clinically negative neck. Five (28% ) of 18 were correctly upstaged by 82%

CT.

The

undetected occult rate to 11%. Friedman and coworkers'7 criteria included lymph node diameter greater than or equal to 1 cm. In Friedman and coworkers' follow-up article on 182 pa¬ tients studied retrospectively, MRI as well as CT was evaluated. One hundred fifty-two patients had preoperative CT. Thirty-two patients had preopera¬ tive MRI. They9 found a statistically significant difference in sensitivity values for physical examination (72% ) compared with CT or MRI (91%, .0006). There were 84 patients who initially presented with a clinically negative neck and 23 (27% ) of 84 pa¬ tients were correctly upstaged by CT or MRI. The undetected occult rate dropped from 39% to 12%. Feinmesser et al18 found that CT was no more sensitive than physical exam¬ ination alone (62% vs 60% ). This was

dropped from 34%

=

based

on a retrospective study on 79 patients who underwent a total of 100 neck dissections. Originally 65 of 100 necks were clinically negative and CT upstaged 20%. Their CT criteria were essentially the same as other authors, with lymph nodes greater than 15 mm in diameter being considered malig¬

nant. A recent article by Close et al10 stud¬ ied 61 patients prospectively and showed CT to have a higher sensitivity

(86%) than physical examination (77%). Computed tomography was found to have a higher overall effi¬ ciency predictive value (81%) than physical examination (69%). This was statistically significant. Thirty-one patients presented with a clinically negative neck and eight of them (26% ) were correctly upstaged with CT. Lydiatt et al11 recently presented a prospective study of 13 patients evalu¬

ated with CT and MRI and found both sensitivity values of 100% compared with 66% for physical ex¬ amination. No evaluation for statisti¬ cal significance was performed. Eight of 13 patients presented with a clini¬ cally negative neck and two (25%) of these eight were correctly upstaged by CT and MRI. The apparent discrepancies in the literature about imaging sensitivities coupled with the inaccuracy of physi¬ cal examination initiated this study to to have

determine whether CT and MRI are useful in the preoperative evaluation for cervical métastases. Our results concur with most of the previously published studies. The sensitivity val¬ ues of 84% and 92% for CT and MRI, respectively, compared favorably with 72% for physical examination. Mag¬ netic resonance imaging had an over¬ all efficiency predictive value of 90% compared with 81% for CT and 78% for physical examination. Other than Feinmesser and coworkers'18 article, there appears to be agreement that CT and now MRI are better predictors of cervical metastasis than physical ex¬ amination alone. The highly encouraging results of MRI obtained in this series may be a function of the subset of our patients and type two statistical error. Contin¬ ued improvement in surface coil design and other imaging parameters, with subsequent improvement in MRI qual¬ ity will likely further increase the use¬ fulness of MRI in evaluation of these patients. With conventional spin echo MRI, central necrosis within lymph nodes can be difficult to detect. Gadolinium-enhanced MRI in con¬ junction with fat suppression tech¬ niques are being developed and should make the identification of central ne¬ crosis easier. Previously cited authors have com¬ mented that CT and MRI have up-

staged the nodal status of the neck preoperatively. Rates of upstaging have ranged from 20% to 55%. This series included 3 (27%) of 11 patients with clinically negative necks that were correctly upstaged by CT or MRI. The preoperative occult rate after CT or MRI dropped to 7%. Computed tomography and MRI may be helpful preoperatively in diag¬ nosing extracapsular spread and ca¬

rotid involvement. The criteria for possible extranodal extension include ill-defined margins around abnormal lymph nodes, edema or thickening of adjacent fat and muscles, and loss of fasciai planes between a mass and ad¬ jacent structure such as the carotid sheath. Magnetic resonance imaging appears better than CT in demonstrat¬ ing whether encased vessels are patent or occluded. Figure 3 demonstrates the loss of tissue planes and carotid artery invasion that was confirmed at the time of resection. Some of these crite¬ ria for extranodal extension are non¬ specific and can be seen secondary to prior radiation or surgery. The results of this study support the use of CT or MRI preoperatively in the evaluation of the clinically negative neck where elective treatment is being considered. These imaging modalities also appear to be beneficial in evaluat¬ ing advanced disease and possibly even the postsurgical/postradiation neck.

References

Strong EW, Schuller DE, Lingeman R, Cox Management of the NO neck. Head Neck Cancer. 1985;1:143-155. 2. Schuller DE, Reiches NA, Haymaker RC, et al. Analysis of disability resulting from treatment including radical neck dissection or modified neck dissection. Head Neck Surg. 1983;6:551\x=req-\ 1.

D.

558.

3. Lee JG, Krause JC. Radical neck dissection: elective, therapeutic and secondary. Ann Otol Rhinol Laryngol. 1975;101:656-659. 4. Rabuzzi DD, Chung CT, Sagerman PH. Prophylactic neck irradiation. Arch Otolaryngol Head Neck Surg. 1980;106:454-455. 5. Nahmun AN, Bane RC, Davidson TM. The case for prophylactic neck dissection. Laryngoscope. 1977;87:588-589. 6. Mancuso AA, Maceri

D, Rice 0, Hanafee W.

CT scan of cervical lymph node cancer. AJR Am J Roentgenol. 1981;136:381-385. 7. Friedman M, Shelton VK, Mafee MM. Meta-

static neck disease: evaluation by CT. Arch Otolaryngol Head Neck Surg. 1984;110:443. 8. Stevens M, Harnsberger MD, Mancuso AA. CT scan of cervical lymph nodes: staging and management of head and neck cancer. Arch Otolaryngol Head Neck Surg. 1985;111:735-739. 9. Friedman M, Mafee MM, Pacella BL, Strorigl TL, Dew LL, Toriumi DM. Rationale for elective neck dissection in 1989. Laryngoscope.

1990;100:54-59. 10. Close L, Merkel M, Vuitch MF, Reisch J,

Schaefer SD. CT evaluation for regional lymph node involvement cancer of oral cavity and oropharynx. Head Neck Surg. 1989;11:309-317. 11. Lydiatt DD, Markin RS, Williams SM, Yonkers AJ. CT and MRI of cervical metastasis. Otolaryngol Head Neck Surg. 1989;101:422. 12. Reede DL, Bergersen RT. The CT evaluation of the normal and diseased neck. Semin CT Ultrasound MR. 1986;7:181-201. 13. Southwick HW. Elective neck dissection for

Downloaded From: http://archotol.jamanetwork.com/ by a Virginia Tech User on 05/14/2015

intraoral cancer. Arch Surg. 1960;80:905-909. 14. Fisch D. Cervical lymphatic system as visualized by lymphangiogram. Ann Otol Rhinol

Laryngol. 1964;73:869-883. 15. Schuller DE, McGuirt WF, McCabe BF, Young D. The prospective significance of metastatic cervical lymph nodes. Laryngoscope. 1980; 90:557-570. 16. noma

Spiro RN, Strong EW. Epidermoid carciof the oral cavity and oropharynx. Arch

Surg. 1973;107:382-384. 17. Spiro HR. Predictive value of tumor thickness

and squamous cell carcinoma confined to the

tongue and floor of the mouth. Am J Surg. 1986;152:345-350. 18. Feinmesser R, Freeman JL, Noyek AM.

Metastatic neck disease: a clinically, radiographically, pathological, correlative study. Arch Otolaryngol Head Neck Surg. 1987;113:1307.