Cervical Lymphadenopathy: SonographicDifferentiation between Tuberculous Nodesand Nodal Metastases from Non–Head andNeck Carcinomas

Michael Ying, MPhil,1 Anil T. Ahuja, FRCR,1 Rhodri Evans, FRCR,1,* Walter King, MD,2

Constantine Metreweli, FRCR1

1 Department of Diagnostic Radiology and Organ Imaging, Prince of Wales Hospital, Shatin,New Territories, Hong Kong2 Department of Surgery, Prince of Wales Hospital, Shatin, New Territories, Hong Kong

Received 15 September 1997; accepted 12 February 1998

ABSTRACT: Purpose. Clinical examination alone can-not differentiate between cervical tuberculous lymph-adenitis and cervical nodal metastases from non–head and neck (NHN) carcinomas because thedistributions of involved lymph nodes are similar. Weevaluated the sonographic features of cervical lymphnodes that could be used to differentiate between the2 categories of nodes.

Methods. We retrospectively reviewed sonogramsof abnormal cervical lymph nodes in 47 patients withproven cervical tuberculous lymphadenitis and in 22patients with proven nodal metastases from NHN car-cinomas.

Results. Abnormal nodes in tuberculous lymphad-enitis and nodal metastases from NHN carcinomaswere commonly found in the supraclavicular fossa(15% and 38%, respectively) and the posterior triangle(70% and 41%, respectively). Statistically significant (p< 0.05) features for differential diagnosis were lymphnodes’ longest diameter, echogenicity, short-to-longaxis ratio, appearance of surrounding soft tissues, andpresence of intranodal cystic necrosis, matting, andposterior enhancement. Nodal size, echogenicity,presence of an echogenic hilum, calcification, coagu-lation necrosis, and sharpness of borders helped inidentifying the abnormal lymph nodes.

Conclusions. Sonographic features that helped todifferentiate between the 2 categories of nodes wereshape, edema of surrounding soft tissue, homogene-

ity, intranodal cystic necrosis, matting, and posteriorenhancement. © 1998 John Wiley & Sons, Inc. J ClinUltrasound 26:383–389, 1998.

Keywords: ultrasonography; cervical lymph nodes; tu-berculosis; infraclavicular carcinomas

The distribution of enlarged lymph nodes is animportant clue in deciding the possible causes

of cervical lymphadenopathy, but when the dis-tribution is similar in different diseases, clinicalexamination alone cannot distinguish betweenthe various causes.1 Sonography of cervical lymphnodes is more accurate than clinical examinationowing to its high sensitivity, but it is limited by itslow specificity; therefore, to increase specificity,sonography is usually combined with fine-needleaspiration biopsy.2 The sonologist, however, mustbe aware of the sonographic features that mayhelp to identify the cause of cervical lymphade-nopathy because fine-needle aspiration biopsy issometimes nondiagnostic and may not be avail-able in all centers. Thus, evaluation of the sono-graphic appearances of abnormal cervical lymphnodes may be helpful in the differential diagnosis.

Tuberculous lymphadenitis is common insoutheast Asia. With the increasing incidence ofacquired immunodeficiency syndrome and associ-ated tuberculosis, it is necessary to make aprompt, early diagnosis. Tuberculous lymphad-enitis is usually found in the supraclavicularfossa and posterior triangle of the neck,3–7 but

*Current address: Radiology Department, Morriston Hospital,Morriston, Swansea SA6 6NL, United KingdomCorrespondence to: A. T. Ahuja

© 1998 John Wiley & Sons, Inc. CCC 0091-2751/98/080383-07

VOL. 26, NO. 8, OCTOBER 1998 383

non–head and neck (NHN) carcinomas, such ascarcinomas of the breast, lung, gastrointestinaltract, and cervix, also commonly metastasize tosupraclavicular lymph nodes.8,9 Because NHNcarcinomas and tuberculous lymphadenitis tendto involve lymph nodes in similar regions, the dis-tribution pattern of diseased lymph nodes cannothelp to narrow the differential diagnosis. The so-nography literature does not discuss the specificfeatures of cervical lymph nodes that help to dif-ferentiate between these 2 conditions. Therefore,the present study evaluated the sonographic fea-tures of cervical lymph nodes that could be usedto differentiate between nodal metastasis fromNHN carcinoma and tuberculous lymphadenitis.

PATIENTS AND METHODS

We retrospectively reviewed sonograms of theneck of 47 patients (17 males and 30 females;mean age, 34 years) with proven tuberculouslymphadenitis and of 22 patients (10 males and12 females; mean age, 50 years) with provennodal metastases from NHN carcinomas. All pa-tients were Asian. All scans were performed withan SSD-650 ultrasound scanner (Aloka, Tokyo,Japan) with a 7.5-MHz linear-array transducerand a 10-MHz sector transducer with a built-inwater bath. Among the 22 patients with NHNcarcinomas, the primary carcinoma was in thebreast in 10 patients, the lung in 11 patients, andthe cervix in 1 patient. None of these patients hadpreviously had radiotherapy to the neck. All pa-tients in both groups had undergone ultrasound-guided fine-needle aspiration biopsy of the largestnode, and if the result was inconclusive, an exci-sional biopsy of the same node had been per-formed. Cytologic and/or histologic investigationsproved all aspirated or biopsied lymph nodes to bemetastases from NHN carcinoma or tuberculouslymphadenitis. In each patient, all other lymphnodes with sonographic features similar to thoseof the aspirated or biopsied node were consideredto be involved by the same pathologic process.

The cervical lymph nodes were classified into 8regions using a method similar to that of Hajek etal11: submental; submandibular; parotid; uppercervical, ie, above the hyoid bone and along thecommon carotid artery and the internal jugularvein; middle cervical, ie, between the hyoid boneand the cricoid cartilage and along the commoncarotid artery and internal jugular vein; lowercervical, ie, below the cricoid cartilage and alongthe common carotid artery and internal jugularvein; supraclavicular fossa; and posterior tri-angle, also known as the accessory chain.

Lymph nodes were assessed using establishedsonographic criteria7,10: distribution, size, shape,presence of an echogenic hilum, internal architec-ture, homogeneity, echogenicity, sharpness ofborders, posterior enhancement, and ancillaryfeatures such as matting and status of surround-ing soft tissues.

Lymph node shape was assessed by the short-to-long axis ratio (S/L). An S/L of less than 0.5indicated an elongated or elliptical node, whereasan S/L of 0.5 or more indicated an oval or roundnode.12

The internal architecture of lymph nodes wasassessed for the presence of calcifications, coagu-lation necrosis, and cystic necrosis. Coagulationnecrosis is an ischemic degeneration found inlymph nodes with malignancy or inflammation.13

Coagulation necrosis is less echogenic than anechogenic hilum and is not continuous with thesurrounding fat.13,14 Areas of coagulation necro-sis tend to be round or oval (ie, S/L ù 0.5),whereas a hilum is usually linear or elliptical (ie,S/L < 0.5).13 Lymph nodes were considered to behomogeneous when they had no calcifications, co-agulation necrosis, or cystic necrosis.

The echogenicity of lymph nodes was deter-mined by comparison with the adjacent muscle(s)and was classified as hypoechoic, isoechoic, or hy-perechoic. Lymph nodes were considered to haveposterior enhancement when the structures pos-terior to the node were more echogenic thanneighboring structures.

Matting was considered to be clumping of mul-tiple abnormal lymph nodes with no normal softtissue between them.

Edema of surrounding cervical soft tissues ap-peared hypoechoic and was associated with a lossof fascial planes. All nodes with edema of adjacentsoft tissue were considered abnormal.

The mean sizes of nodal metastases from NHNcarcinomas and tuberculous lymphadenitis werecompared using Student’s t-test. The sonographicfeatures of tuberculous lymphadenitis and nodalmetastases from NHN carcinomas were com-pared using the chi-squared test. A p value of lessthan 0.05 was considered statistically significant.

RESULTS

A total of 315 lymph nodes were involved by tu-berculosis, and 114 lymph nodes were involved bymetastatic NHN carcinoma. Involved nodes weremost commonly found in the supraclavicular fossa(15% of tuberculous nodes and 38% of NHN me-tastases) and the posterior triangle (70% of tuber-

YING ET AL

384 JOURNAL OF CLINICAL ULTRASOUND

culous nodes and 41% of NHN metastases) (Fig-ure 1).

The size of lymph nodes (long axis × short axis)involved by tuberculosis ranged from 9 × 3 mm to42 × 20 mm, and nodal metastases ranged from 7× 5 mm to 42 × 33 mm. The mean short axis of thelymph nodes involved by tuberculosis (10.2 mm)was not significantly different than that of nodalmetastases (10.0 mm) (p > 0.05). However, thedifference in the mean long axis of lymph nodesinvolved by tuberculosis (19.0 mm) and that fornodal metastases (15.6 mm) was statistically sig-nificant (p < 0.05).

Table 1 shows the numbers, sizes, and sono-graphic features of lymph nodes involved bymetastatic NHN carcinoma and those involved bytuberculosis and gives the significance of differ-ences between the 2 diseases.

The significant sonographic features for differ-ential diagnosis included echogenicity, sharpnessof border, shape, edema of surrounding soft tis-sue, homogeneity, intranodal cystic necrosis, mat-ting, and posterior enhancement. The majority oflymph nodes involved by either tuberculosis ormetastatic NHN carcinoma were hypoechoic(100% and 92%, respectively) and had sharp bor-ders (62% and 74%, respectively). Therefore, al-though the difference in echogenicity was statis-tically significant, it was not large enough to beclinically useful. Only 5% of nodal metastases hadan S/L of less than 0.5, whereas 21% of tubercu-lous nodes had an S/L of less than 0.5 (Figure 2).Tuberculous lymphadenitis had a higher inci-dence of abnormal surrounding soft tissues (49%),cystic necrosis (60%) (Figures 2 and 3), and mat-

ting (59%) (Figure 4) than did nodal metastases(4%, 8%, and 8%, respectively). Heterogeneouslymph nodes were commonly found in tuberculouslymphadenitis (64%), whereas nodal metastaseswere predominantly homogeneous (88%). Poste-rior enhancement was more common in tubercu-lous lymphadenitis (22%) than in nodal metasta-ses (4%) (Figure 3).

The sonographic features that were insignifi-cant to the differential diagnosis included thepresence of an echogenic hilum, intranodal calci-fications, and intranodal coagulation necrosis.Lymph nodes involved by either tuberculosis ormetastatic NHN carcinoma commonly had no hi-lum (86% and 87%, respectively). Calcificationsand coagulation necrosis were rarely found inlymph nodes involved by tuberculosis (1% and6%, respectively) or in nodal metastases (0% and7%, respectively).

DISCUSSION

Cervical tuberculous lymphadenitis and cervicalnodal metastases from NHN carcinomas cannotbe differentiated solely on the basis of the distri-bution of involved lymph nodes. In this study, themajority of lymph nodes involved by tuberculosiswere found in the supraclavicular fossa and theposterior triangle, consistent with previous re-ports.3–7 However, these were also the most com-mon sites of NHN metastases, in contrast to pre-vious reports in which only supraclavicularlymph nodes were involved by metastases.8,9 Thisdifference is probably due to the fact that in this

FIGURE 1. Schematic of the neck showing the distribution of tuberculous lymph nodes (white bars) and nodal metastases from non–head and neckcarcinomas (black bars).

CERVICAL LYMPHADENOPATHY

VOL. 26, NO. 8, OCTOBER 1998 385

study, patients with NHN carcinoma were in-cluded regardless of whether they had abnormalsupraclavicular lymph nodes, whereas previousstudies included only patients who had abnormalsupraclavicular lymph nodes, which resulted inselection bias.

The sonographic features of cervical tubercu-lous lymphadenitis and cervical nodal metastasesfrom NHN carcinomas can be grouped into threegeneral categories: (1) features that indicated ab-normality but were not specific, (2) features thatfavored metastasis from NHN carcinoma, and (3)features that favored tuberculous lymphadenitis.The sonographic features of nodal size, echo-genicity, presence of an echogenic hilum, calcifi-cation, coagulation necrosis, and sharpness ofborders were generally not helpful in the differ-

ential diagnosis. Although previous studiesshowed nodal size to be of little value in differen-tiating benign from malignant nodes,15,16 wefound a statistically significant difference in themean long axes of lymph nodes involved by tuber-culosis and those involved by metastasis (p <0.05). Unfortunately, this difference was not largeenough to be clinically useful.17 Echogenicity wasunhelpful because the majority of lymph nodesinvolved by either disease were hypoechoic. Like-wise, the presence of an echogenic hilum was nothelpful because the majority of lymph nodes hadno hilum. Calcification in nodal metastasis is gen-erally rare, except in nodal metastases from pap-illary carcinoma of the thyroid.18,19 Calcificationmay be found in tuberculous lymphadenitis, but itis not pathognomonic for tuberculosis.5 In this

TABLE 1

Sonographic Features of 315 Tuberculous Lymph Nodes and 114 Nodal Metastases from

Non–Head and Neck Carcinomas

Sonographic Feature

Number of Lymph Nodes (%)

Significance of Differencebetween the 2 Diseases

TuberculousLymphadenitis

NodalMetastasis

SizeRange, mm(long axis × short axis) 9 × 3 to 42 × 20 7 × 5 to 42 × 33Mean short axis, mm 10.2 10.0 NSMean long axis, mm 19.0 15.6 p < 0.05

Echogenicity*

Hypoechoic 315 (100%) 105 (92%) p < 0.0001Isoechoic 0 (0%) 9 (8%)

HilumPresent 45 (14%) 15 (13%) NSAbsent 270 (86%) 99 (87%)

CalcificationPresent 4 (1%) 0 (0%) NSAbsent 311 (99%) 114 (100%)

Coagulation necrosisPresent 18 (6%) 8 (7%) NSAbsent 297 (94%) 106 (93%)

Nodal borderSharp 195 (62%) 84 (74%) p < 0.03Not sharp 120 (38%) 30 (26%)

ShapeS/L < 0.5 67 (21%) 6 (5%) p < 0.0001S/L ù 0.5 248 (79%) 108 (95%)

Surrounding tissuesNormal 160 (51%) 110 (96%) p < 0.0001Abnormal 155 (49%) 4 (4%)

HomogeneityHomogeneous 112 (36%) 100 (88%) p < 0.0001Heterogeneous 203 (64%) 14 (12%)

Cystic necrosisPresent 188 (60%) 9 (8%) p < 0.0001Absent 127 (40%) 105 (92%)

MattingPresent 185 (59%) 9 (8%) p < 0.0001Absent 130 (41%) 105 (92%)

Posterior enhancementPresent 68 (22%) 5 (4%) p < 0.0001Absent 247 (78%) 109 (96%)

Abbreviations: NS, not significant (p ù 0.05); S/L, short-to-long axis ratio.*No nodes in either group were hyperechoic.

YING ET AL

386 JOURNAL OF CLINICAL ULTRASOUND

study, calcification was not a useful distinguish-ing feature because the majority of lymph nodeshad no calcification. Similarly, coagulation necro-sis occurred too infrequently to be useful. Finally,previous reports have described nodal metastasesas having sharp, well-defined borders and inflam-matory lymph nodes as having poorly defined bor-ders.19,20 In this study, the majority of tubercu-lous lymph nodes and nodal metastases hadsharp borders. Although sharpness of borders

was found to be a statistically significant feature,it is unlikely to be clinically useful.

Sonographic features that favored metastasisfrom NHN carcinoma were normal surroundingsoft tissues, absence of intranodal cystic necrosis,absence of matting, and homogeneity. Comparedwith tuberculous lymphadenitis, almost twice asmany nodal metastases had normal surroundingsoft tissues, and more than twice as many showedno intranodal cystic necrosis and/or were homo-

FIGURE 3. Longitudinal sonogram of the posterior triangle of the neck showing tuberculous lymph nodes withintranodal cystic necrosis (white arrows). Note the enhancement distal to the lymph nodes (black arrows).

FIGURE 2. Longitudinal sonogram of an oval lymph node in the posterior triangle of the neck with extensiveintranodal cystic necrosis (black arrows). This appearance is suggestive of tuberculous lymphadenitis.

CERVICAL LYMPHADENOPATHY

VOL. 26, NO. 8, OCTOBER 1998 387

geneous. Matting of lymph nodes, which was seenin 59% of tuberculous lymph nodes, occurred inonly 8% of nodal metastases.

Sonographic features that favored tuberculouslymphadenitis were more striking. These were anS/L of less than 0.5, abnormal surrounding softtissues, heterogeneity, intranodal cystic necrosis,matting, and posterior enhancement. In agree-ment with van Overhagen et al,21 who reportedthat the S/L of malignant nodes was greater thanthat of benign nodes, we found that an S/L of lessthan 0.5 occurred more often in tuberculouslymphadenitis than in nodal metastases. We alsofound that 49% of tuberculous lymph nodesshowed abnormal surrounding soft tissues, whichincluded edema or induration of the subcutaneoustissue and/or adjacent muscle; only 4% of nodalmetastases showed this feature. The high inci-dence of heterogeneity seen in tuberculouslymphadenitis (64%) probably resulted from thehigh incidence of intranodal cystic necrosis (60%).Intranodal cystic necrosis is common in tubercu-lous lymphadenitis and can also be identified byCT, on which it appears as a central low-densityarea with rim enhancement.22 While intranodalcystic necrosis may occur in malignant lymphnodes,23 such an occurrence in our study was in-frequent. Matting of lymph nodes is another com-mon feature in tuberculous lymphadenitis,6,24

and in cases of tuberculous lymphadenitis inwhich there has been considerable periadenitis,the involved lymph nodes will be matted togetherwith fibrous tissue.3 Finally, except for lymphnodes affected by non-Hodgkin’s lymphoma,17,25

posterior enhancement is not frequently seen innodal metastases because they are composed oftumor cells separated by supportive tissue thatproduces numerous interfaces. Only 4% of thenodal metastases in this study showed posteriorenhancement. However, 22% of tuberculouslymph nodes showed this feature, an incidencethat again may be due to the high incidence ofintranodal cystic necrosis.

Our study shows that of all the sonographiccriteria previously described for abnormal lymphnodes, the only ones likely to help differentiatebetween tuberculous lymphadenitis and nodalmetastasis from NHN carcinoma are an S/L ofless than 0.5 versus an S/L of 0.5 or more, abnor-mal versus normal surrounding soft tissue, het-erogeneity versus homogeneity of the node, andpresence versus absence of intranodal cystic ne-crosis, matting, and posterior enhancement.

REFERENCES

1. Lindberg R: Distribution of cervical lymph nodemetastases from squamous cell carcinoma of theupper respiratory and digestive tracts. Cancer1972;29:1446.

2. Baatenburg de Jong RJ, Rongen RJ, Lameris JS, etal: Metastatic neck disease: palpation vs ultra-sound examination. Arch Otolaryngol Head NeckSurg 1989;115:689.

3. Newcombe J: Tuberculosis. Tuberculous glands inthe neck. Br J Hosp Med 1979;22:553.

4. Appling D, Miller RH: Mycobacterial cervicallymphadenopathy: 1981 update. Laryngoscope1981;91:1259.

5. Reede DL, Som PM: Lymph nodes. In Som PM,

FIGURE 4. Longitudinal sonogram of the posterior triangle of the neck showing multiple hypoechoic tuber-culous lymph nodes matted together without normal intervening soft tissue.

YING ET AL

388 JOURNAL OF CLINICAL ULTRASOUND

Bergeron RT, eds: Head and Neck Imaging, 2ndedn. St. Louis, MO, Mosby–Year Book, 1991, p.558.

6. Reede DL, Bergeron RT: Cervical tuberculous ad-enitis: CT manifestations. Radiology 1985;154:701.

7. Ahuja A, Ying M, Evans R, et al: The application ofultrasound criteria for malignancy in differentiat-ing tuberculous cervical adenitis from metastaticnasopharyngeal carcinoma. Clin Radiol 1995;50:391.

8. Cervin JR, Silverman JF, Loggie BW, et al: Vir-chow’s node revisited. Analysis with clinicopatho-logic correlation of 152 fine-needle aspiration biop-sies of supraclavicular lymph nodes. Arch PatholLab Med 1995;119:727.

9. Kiricuta IC, Willner J, Kolbl O, et al: The prognos-tic significance of the supraclavicular lymph nodemetastases in breast cancer patients. Int J RadiatOncol Biol Phys 1994;28:387.

10. Ishii J, Amagasa T, Tachibana T, et al: Ultrasonicevaluation of cervical lymph node metastasis ofsquamous cell carcinoma in oral cavity. Bull TokyoMed Dent Univ 1989;36:63.

11. Hajek PC, Salomonowitz E, Turk R, et al: Lymphnodes of the neck: evaluation with US. Radiology1986;158:739.

12. Tohnosu N, Onoda S, Isono K: Ultrasonographicevaluation of cervical lymph node metastases inesophageal cancer with special reference to the re-lationship between the short to long axis ratio (S/L)and the cancer content. J Clin Ultrasound 1989;17:101.

13. Evans RM, Ahuja A, Metreweli C: The linear echo-genic hilus in cervical lymphadenopathy—a sign ofbenignity of malignancy? Clin Radiol 1993;47:262.

14. Sakai F, Kiyono K, Sone S, et al: Ultrasonic evalu-ation of cervical metastatic lymphadenopathy. JUltrasound Med 1988;7:305.

15. Solbiati L, Arsizio B, Rizzatto G, et al: High-resolution sonography of cervical lymph nodes inhead and neck cancer: criteria for differentiation of

reactive versus malignant nodes. Radiology 1988;169(P):113.

16. Vassallo P, Wernecke K, Roos N, et al: Differentia-tion of benign from malignant superficial lymph-adenopathy: the role of high-resolution US. Radi-ology 1992;183:215.

17. Bruneton JN, Normand F: Cervical lymph nodes.In Bruneton JN, ed: Ultrasonography of the Neck.Berlin, Springer-Verlag, 1987, p. 81.

18. Ahuja AT, Chow L, Chick W, et al: Metastatic cer-vical nodes in papillary carcinoma of the thyroid:ultrasound and histological correlation. Clin Ra-diol 1995;50:229.

19. Som PM: Lymph nodes of the neck. Radiology1987;165:593.

20. Shozushima M, Suzuki M, Nakasima T, et al: Ul-trasound diagnosis of lymph node metastasis inhead and neck cancer. Dento-Maxillo-Facial Radi-ology 1990;19:165.

21. van Overhagen H, Lameris JS, Berger MY, et al:Supraclavicular lymph node metastases in carci-noma of the esophagus and gastro-esophageal junc-tion: assessment with CT, US, and US-guided fine-needle aspiration biopsy. Radiology 1991;179:155.

22. Pombo F, Rodriguez E, Mato J, et al: Patterns ofcontrast enhancement of tuberculous lymph nodesdemonstrated by computed tomography. Clin Ra-diol 1992;46:13.

23. Rubaltelli L, Proto E, Salmaso R, et al: Sonographyof abnormal lymph nodes in vitro: correlation ofsonographic and histologic findings. AJR Am JRoentgenol 1990;155:1241.

24. Patel MP, Adler L, Alexander LL: The value of CTscanning in the diagnosis of tuberculous lymphad-enitis of the neck. Computerized Radiology 1984;8:21.

25. Bruneton JN, Roux P, Caramella E, et al: Ear,nose, and throat cancer: ultrasound diagnosis ofmetastasis to cervical lymph nodes. Radiology1984;152:771.

CERVICAL LYMPHADENOPATHY

VOL. 26, NO. 8, OCTOBER 1998 389