Mediastinal staging of non-small cell lung carcinoma using computed and positron-emission tomography.

ABSTRACT

Background. We evaluated the accuracy of computed tomography (CT) and positron-emission tomography (PET) in the mediastinal staging of non-small cell lung cancer.

Methods. Between May 14, 1999, and November 28, 2000, computerized tomography (CT) and positron-emission tomography (PET) were used to clinically stage 94 consecutive patients with non-small cell carcinoma of the lung (NSCCL). All patients underwent subsequent surgical staging with mediastinoscopy, anterior mediastinotomy, and/or thoracotomy with mediastinal lymphadenectomy.

Results. Overall accuracy was the same for both procedures. False-negative results occurred 3 times more often with CT; false-positive results occurred twice as often with PET. Sensitivity and specificity were 64% and 94%, respectively, for CT, versus 88% and 86%, respectively, for PET. Positive and negative predictive values were 80% and 88%, respectively, for CT, versus 71% and 95%, respectively, for PET.

Conclusion. In addition to routine use of CT, PET seems to achieve high negative predictive value in the evaluation of mediastinal disease; PET seems particularly helpful in assessing absence of tumor in bulky nodes after neoadjuvant chemotherapy and/or radiotherapy.

**********

IT HAS BEEN STATED, (1,2) "The tumor-node-metastasis system (of staging) has proved to be the best method ... to determine treatment strategies and ultimate prognosis ... (and) the involvement of local (intrapulmonary) or regional (extrapulmonary) lymph nodes remains the most important predictor of outcome after resection of lung cancer."

Regional, extrapleural, and distant metastatic disease so adversely affect the prognosis of patients with non-small cell carcinoma of the lung (NSCCL) that surgical resection is not an appropriate option for treatment in most patients with stage III and IV disease (ie, patients with regional mediastinal or distant spread of disease, respectively), at least not until they have received preoperative therapy.

Computed tomography (CT) has supplanted simple chest roentgenograms, planigrams, fluoroscopy, and radionuclide bone scanning for routine clinical staging purposes. Now the emergence of positron-emission tomography (PET) allows a marriage of metabolic imaging (PET) to complement the anatomic imaging of CT, hopefully to more appropriately guide therapy.

Positron-emission tomography is a relatively noninvasive nuclear technique that delivers high resolution images using positron-emitting radiotracers, most commonly 2-(F-18)-fluoro-2-dioxy-D-glucose (FDG). Because malignant cells typically exhibit accelerated glycolysis, FDG is rapidly concentrated and well imaged, with a high tumor-to-background ratio. There is also concentration in the brain, urinary tract, and areas of inflammation.

Positron-emission tomography may be used as a diagnostic aid in evaluating pulmonary nodules/masses, extrathoracic staging, and monitoring therapy, as well as for mediastinal staging, which is the subject of this study.

MATERIALS AND METHODS

Between May 14, 1999, and November 28, 2000, all 104 patients evaluated for surgical resection for non-small cell carcinoma of the lung (NSCCL) were evaluated with CT and PET. Computed tomography scans varied by manufacturer, model, institution, and slice thickness (5 to 10 mm); all were conducted with intravenous contrast. Whole body PET scans were all done with an ECAT model 951/31 Siemens-CTI (Siemens AG, Munich, Germany) with an axial field view of 10.8 cm and a full-width at half-maximal resolution of 6 mm.

Computed tomography scans were interpreted by the radiologist of record, the attending pulmonary medicine specialist, and the surgeon involved. If any of these 3 principals deemed the mediastinum positive or worrisome (or in doubt) for lymphadenopathy, the CT scan was classified as clinically positive (or indeterminate) for mediastinal spread of disease. Lymphadenopathy of the mediastinum was not solely defined by short-axis length / width of nodes. Rather, the whole image on the CT, including the primary tumor, its size and location, and the appearance of the mediastinum, as well as the clinicians' impressions of the appearance and performance status of the patient, were considered in declaring the likely presence of mediastinal disease (positive for N2 disease). Disease spread to the mediastinum was to be either positive or negative according to clinical suspicion (preoperative) or pathologic fact (postoperative).

All patients had surgical excision of their lung tumor if it was benign. If it was malignant, they had surgical staging by mediastinoscopy, anterior mediastinotomy, and/or thoracotomy for primary lung resection and thorough mediastinal lymphadenectomy. All surgery was done by one surgeon. The Figure illustrates the definitions and calculations used to compare results of CT and PET. We use this experience to examine the utility of CT and PET in the staging of N2 disease in patients with NSCCL.

RESULTS

Of the 104 patients, 10 were found to have benign disease: 3 granulomas, 2 subpleural lymph nodes, and 1 patient each with hamartoma, abscess, typical carcinoid, Schwannoma, and localized bronchiolitis obliterans with organizing pneumonia (BOOP). Computed tomography did not delineate mediastinal metastases in any of these 10 cases, while PET erroneously indicated positive results in 1 case that was found to have no mediastinal metastases on pathologic examination.

Of the 94 patients with NSCCL, the primary tumors were worrisome for malignancy or indeterminate by CT in all cases. Positron-emission tomography was misleading in 4 cases, denoting benign disease, yet pathologically all 4 were adenocarcinomas with predominant bronchoalveolar features. Of 4 other patients with multicentric tumors, CT accurately predicted the number of separate malignancies in 2 (50%), PET in none.

This group of 94 patients with NSCCL forms the cohort of our study in examining the utility of CT and PET in preoperative recognition of mediastinal (N2) spread of disease. The presence or absence of likely mediastinal disease could not be characterized in 2 patients evaluated by CT and 6 patients evaluated by PET (Table 1). The most likely explanations for this were either the paratracheal and/or hilar locations of their primary tumors (CT) or misinterpretation of whether hilar or mediastinal lymph nodes were imaged (PET). Of the remaining patients, the sensitivity in detecting N2 disease was 64% for CT and 88% for PET. The specificity in detecting the absence of N2 disease in these patients was 94% for CT, whereas for PET it was 86%. The positive and negative predictive values for CT were 80% and 88%, respectively, while for PET they were 71% and 95%, respectively. The overall accuracy of both modalities was 86%, excluding those cases judged to be indeterminate (Table 1).

Table 2 depicts the utility of CT and PET in delineating the presence or absence of mediastinal spread of disease in 8 patients treated with preoperative (neoadjuvant) chemotherapy and/or radiation therapy. For these patients, the sensitivity of both techniques was equal and marginal (50%), but PET had higher specificity, positive and negative predictive values, and overall accuracy.

DISCUSSION

Computed tomography resulted in 3 times as many false-negative results as PET regarding mediastinal status, whereas PET resulted in about twice as many false-positive results as CT (Table 1). While PET and CT were equally accurate (86%) in assessing untreated disease, PET outperformed CT in our small sample of patients treated neoadjuvantly (Table 2).

Comparing our results to 3 reports in the literature reveals a good amount of variability, with the exception of the negative predictive value of PET (Table 3). (3-5) The negative predictive value of PET consistently ranged from 93% to 95% among the 3 studies cited, which is reassuring when using PET to help select patients for resection.

It is difficult to know why the sensitivity of CT was so low in the experiences of Saunders et al. (3) Perhaps they narrowly assessed mediastinal adenopathy, without context of primary tumor size and/or location relative to the mediastinum. The experience of Pieterman et al (4) with PET in the Netherlands closely matches our own results.

The European experience (5) with minimal staging is cited in Table 3. Despite their results, mediastinoscopy/mediastinotomy does not assess subcarinal anatomy. The subcarinal area is precisely where noninvasive testing with methods such as CT and PET has great theoretical advantage. The advantage of combining an anatomic diagnostic test (CT) with PET, which depicts metabolically active nodules, seems to be borne out by others, (3,4,6) as well as our own results in which PET reduced the CT rate of false-negative results by two thirds.

We do not recommend using CT and/or PET in lieu of mediastinoscopy and/or anterior mediastinotomy in all cases. Similarly, we have not yet concluded that both CT and PET should routinely be used in all cases of suspected NSCCL. We are comfortable, when progressing directly to thoracotomy after negative CT and PET, performing thorough mediastinal lymphadenopathy at the time of resection of the primary tumor. We further recognize that others have shown that the routine use of PET after negative CT may be financially efficacious as well. (6)

For N2 disease in patients with NSCCL, directly proceeding to resection after negative CT and PET seems appropriate. Clinical appearance of disease may be misleading, however. (7) Watanabe (8) noted his institution's "true-positive rate of (N2 disease) was only 57% in patients pathologically evaluated post-resection following preoperative 'positive' CT scanning over the years 1980-1990." That same study noted that if all patients with the suggestion of N2 disease on CT were excluded from surgery, 39% of the ultimately resectable pathologic stage N0 and N1 patients would have been denied surgery. We agree with the opinion that "one does not cure NSCCL by unnecessarily excluding patients from surgery." (9)

Further, in our experience (10) and that of others, (11,12) minimal N2 disease has clearly been shown not to be associated with the devastating prognosis of bulky N2 disease. The 5-year survival rate of 42% in our patients with occult N2 disease (10) is matched by results reported from Memorial Sloan-Kettering (11) and Tufts, (12) in contradistinction to the vast majority of patients with stage N2 disease (90%) who have bulky disease, in whom the 3-year survival rate may be less than 10%. (11)

CONCLUSIONS

Non-small cell carcinoma of the lung is the leading cause of death from cancer. Refinements are being made in diagnostic tests in an attempt to reliably sort out patients with local, resectable disease from those with regional and distant metastatic disease, where optimum treatments may be neoadjuvant therapy or chemotherapy and radiotherapy with curative intent, respectively.

Based on our results, as well as others cited, (3,4) we believe that the addition of PET scanning to routine use of CT is progress in the direction of less invasive and more efficacious diagnosis and therapy. In our experience and the experience of others, (3,4) PET scanning seems to achieve a high negative predictive value (93% to 95%) in the evaluation of mediastinal disease. Positron-emission tomography seems to be particularly helpful in evaluating patients whose bulky nodes may persist on CT after chemotherapy and/or radiotherapy. Metabolically sensitive PET seems to achieve higher overall accuracy (88% vs 63%) in determining the persistence/absence of viable cancer after such therapy.

          Test Result       Mediastinal Pathology

          Positive          Negative

Positive  True positive  A  B False positive

Negative  False negative C  D True negative

Definitions and calculations used to compare results of CT and PET:

Sensitivity = accuracy in detecting individuals with disease = A/(A+C).

Specificity = accuracy in detecting individuals without disease =
D/(B+D).

Positive predictive value = probability that disease is present when
test is positive = A/(A+B).

Negative predictive value = probability that disease is absent when test
is negative = D/(C+D).

Overall accuracy = percent correct classification of all individuals =
(A+D)/(A+B+C+D).

Scientific assessment according to documented pathology.
TABLE 1

Mediastinal Status in Non-Small Cell Carcinoma of the Lung

                          Mediastinal Pathology

                     Positive          Negative  Total

CT Results

    Positive            16                 4      20
    Negative             9                63      72

    Total *             25                67      92

PET Results

    Positive            22                 9      31
    Negative             3                54      57

    Total +             25                63      88
                                     Positive    Negative
P                                   Predictive  Predictive
Modality  Sensitivity  Specificity    Value       Value     Accuracy

CT            64%          94%         80%         88%        86%
PET           88%          86%         71%         95%        86%

* cases had indeterminate status.

+ 6 cases had indeterminate status.

CT = Computed tomography

PET = positron-emission tomography.
TABLE 2

Mediastinal Status in Non-Small Cell Carcinoma of the Lung After
Neoadjuvant Therapy

                          Mediastinal Pathology

                     Positive          Negative  Total

CT Results

    Positive             1                2        3
    Negative             1                4        5

     Total               2                6        8

PET Results

    Positive             1                0        1
    Negative             1                6        7

     Total               2                6        8
                                     Positive    Negative
                                    Predictive  Predictive
Modality  Sensitivity  Specificity    Value       Value     Accuracy

CT            50%          67%          33%        80%        63%
PET           50%         100%         100%        86%        88%

CT = Computed tomography, PET = positron-emission tomography.
TABLE 3

Comparison of CT and PET in Assessing the Presence or Absence of
Mediastinal Pathology in Several Studies

                                      Study

           Hospital    Kiernan et al  Saunders et al (3)
           Modality      (N = 94)        (N = 84)

CT
          Sensitivity       64%            20%
          Specificity       94%            90%
          +PV               80%            30%
          -PV               88%            84%
PET
          Sensitivity       88%            71%
          Specificity       86%            97%
          +PV               71%            86%
          -PV               95%            93%

                            Study

          Pieternan et al (4)
             (N = 102)         Dillemans et al (5)

CT
                75%                  72%
                66%                 100%
                NA                    -
                NA                    -
PET
                91%                  72%
                86%                 100%
                74%                   -
                95%                   -

+ PV = Positive predictive value, - PV = negative predictive value, NA =
not applicable.

References

(1.) Luketich JD, Ginsberg RJ: Diagnosis and staging of lung cancer. Lung Cancer. Johnson BE, Johnson DH (eds). New York, Wiley-Liss, 1995, pp 161-173

(2.) Mountain CF: Revisions in the International System for Staging Lung Cancer. Chest 1997; 111:1710-1723

(3.) Saunders CAB, Dussek JE, O'Doherty MJ, et al: Evaluation of fluorine-18-fluorodeoxyglucose whole body positron emission tomography imaging in the staging of lung cancer. Ann Thorac Surg 1999;67:790-797

(4.) Pieterman RM, van Putten JWG, Meuzelaar JJ, et al: Preoperative staging of non-small cell lung cancer with positron-emission tomography. N Engl J Med 2000;343:254-261

(5.) Dillemans B, Deneffe G, Verschakelen J, et al: Value of computed tomography and mediastinoscopy in preoperative evaluation of mediastinal nodes in non-small cell lung cancer. Eur J Cardiothorac Surg 1994;8:37-42

(6.) Scott WJ, Shepherd J, Gambhir SS: Cost effectiveness of FDG-PET for staging non-small cell lung cancer: a decision analysis. Ann Thorac Surg l998; 66:1876-1885

(7.) Whittlesey D: Prospective computed tomographic scanning in the staging of bronchogenic carcinoma. J Thorac Cardiovasc Surg 1988; 95:876-882

(8.) Watanabe Y, Shimizu J, Oda M, et al: Aggressive surgical intervention in N2 non-small cell cancer of the lung. Ann Thorac Surg 1991; 51:253-261

(9.) Pairolero PC, Miller D: Results of surgical resection at Fair-fax Hospital. Presented at INOVA Fairfax Conference on Non-Small Cell Cancer of the Lung, Jan 2001, Fairfax, Va

(10.) Kiernan PD, Sheridan M, Byrne WD, et al: Stage II and IIIA non-small cell cancer of the lung: results of surgical resection at Fairfax Hospital 1994; Va Med Q 1994; 121:172-178

(11.) Martini N, Flehinger BJ: The role of surgery in N2 lung cancer. Surg Clin North Am 1987; 67:1037-1049

(12.) Daly BT, Mueller JD, Faling LJ, et al: N2 lung cancer: outcome in patients with false-negative computed tomographic scans of the chest. J Thorac Cardiovasc Surg 1993; 105:904-911

RELATED ARTICLE: KEY POINTS

* Overall accuracy for computed tomography (CT) and positron-emission tomography (PET) is approximately the same--in our experience, 86%.

* False-negatives occurred three times more commonly with CT; false-positives occurred two times more often with PET.

* Positron-emission tomography seems to achieve a high negative predictive value (93% to 95%) in the evaluation of mediastinal disease.

* Positron-emission tomography seems to be particularly helpful in evaluating patients with bulky nodes that may persist after chemotherapy and/or radiation therapy as delivered neoadjuvantly.

From the Sections of Thoracic Surgery and Pulmonary Medicine, and the Department of Medicine, Inova Fairfax and Inova Alexandria Hospitals, Fairfax and Alexandria, Va.

Reprint requests to Paul Kiernan, MD, 3301 Woodburn Rd. Suite 301, Annandale, VA 22003.