Comparison of Immunoscintigraphy and Computerized ... · [CANCER RESEARCH 51. 5704-5711. October 15, 1991] Comparison of Immunoscintigraphy and Computerized Tomography in Identifying
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
[CANCER RESEARCH 51. 5704-5711. October 15, 1991]
Comparison of Immunoscintigraphy and Computerized Tomography in IdentifyingColorectal Cancer: Individual Lesion Analysis1
Raffael M. Corbisiero, Dave M. Yamauchi, Lawrence E. Williams, Jose M. Esteban, Tamara Odom-Maryon,and J. David Bearty2
Departments of General Oncologie Surgery [R. M. C., J. D. B.], Diagnostic Radiology [D. M. Y., L. E. W.], Anatomic Pathology [J. M. E.J, and BiostatisticsIT. O-MJ, City of Hope National Medical Center, Duarte, California 91010
ABSTRACT
Monoclonal antibody scintigraphy with "'In-ZCE025 was used in
IS3 with radiolabeled monoclonal antibody specific for a
tumor marker has shown potential value in cancer patientmanagement (1-6). In colorectal adenocarcinoma, presurgicalimaging has confirmed known sites of metastatic disease andhas identified occult sites of disease that had not been localizedpreviously (7-11). The sensitivity of IS for abdominal and pelvicdisease has been reported to vary over a wide range, i.e., from38% (12) to >90% (11), depending upon the antibody, theradionuclide, the disease location, and, most importantly, themethod of scoring sensitivity (lesion versus region analysis).Similarly, CT has had sensitivity for abdominal metastaticdisease ranging from 30% (2, 4) to >70% (13).
rotomy was planned. In these studies, IS and disease statuswere compared regionally in the colorectum, liver, extrahepaticabdomen, and all extraabdominal areas.
In the present study, IS and CT were compared with diseasestatus on a lesion by lesion basis. Patients had preoperative ISwith '"In-ZCE025 and CT, followed by abdominal exploration
and pathological evaluation of disease sites. The sensitivity andaccuracy of IS for CEA-specific identification and localizationof individual sites of intraabdominal colorectal cancer wascompared with CT for nonspecific identification and localization of the same individual lesions.
MATERIALS AND METHODS
Antibody Preparation and Administration. A murine IgGl intactmonoclonal antibody specific for CEA (ZCE025) was conjugated withdiethylenetriamine pentaacetic acid (14) and labeled with '"In (doserange 2.5-7.1 mCi) using a kit provided by Hybritech Inc. (San Diego,CA). Labeling efficiency was confirmed to be 80% or greater by thinlayer chromatography. Labeled '"In-ZCE025 (1 mg) was mixed with
either 19 or 39 mg of unlabeled ZCE025 in approximately 50 ml ofsaline to produce the final "'In-ZCE025 solution which was injectedi.v. during a 5- to 6-min period.
Imaging. Preoperative IS was performed at various times between 48and 192 h following injection of'"In-ZCE025. In all patients, anterior
and posterior (whole body or regional) planar scans were obtained.Prior to February 1988, a Technicare Omega 500 camera was utilized;after this date, a Toshiba 901 camera which has SPECT capability wasutilized. Scintiscans used windows centered over the '"In photon
energies of 172 and 247 keV. Tomographie images of the liver (SPECT)were routinely obtained in 3 planes. Other sites were selected forSPECT on the basis of the planar images or other clinical data.
Preoperative axial CT scans of the abdomen were obtained using aSynerview 1200 SX (Picker International). Chest and/or pelvis CTscans were also performed whenever possible. Each of the IS and CTscans was reviewed retrospectively for this study by an experienceddiagnostic and nuclear medicine radiologist (D. Y.) and a surgeon (R.C.).
Surgery. Abdominal exploration was performed in each case. Duringsurgery, accessible regions of the abdomen and pelvis were inspected
with particular attention to areas of abnormality on IS scan and CTscan. Biopsy and resection of tumor and adjacent structures wereundertaken as medically indicated in the best interests of the patient.Tissues were analyzed by gross visual and microscopic pathologicalexamination. Size measurements of some resected lesions were available from the gross pathological assessment. Content of'"In in biopsy
or resection tissue was measured by weighing a sample and counting iton a gamma counter (GammaTrac 1193; TM Analytic, Elk Grove, IL),expressing the results, corrected for physical decay of '"In, as the %
ID/g. This latter parameter is a measure of the density of antibodywithin a tissue, which consequently reflects how visible it is usinggamma camera scintigraphy.
Patient Selection. Between July 1986 and January 1990, all patientsstudied by IS using '"In-ZCE025 were eligible for consideration for
this retrospective review. Patients included all had colorectal carcinoma(primary, hepatic metastasis, or known or suspected recurrence) andabdominal exploration. The '"In-ZCE025 injection for IS was per
formed within 2 weeks prior to surgery, and the preoperative CT scanswere performed within 6 weeks prior to "'In-ZCE025 injection. The
median time separation between the CT scan and operation was 19days (range, 1-69 days). Greater than 80% of the patients had their CTscans performed within 5 weeks of surgery. Except for one patient (69days), all had the CT scan performed within 8 weeks. This time intervalwas appropriate for retrospective comparison of operating roomfindings.
Lesion Analysis. Any area considered to represent colorectal cancerby any one of IS, CT, surgery, or pathology was defined as a lesion. OnIS scan, any discrete area of relatively increased '"In uptake ("hot")
was defined as a lesion. Discrete areas of decreased uptake (i.e., coldor photopenic "lesions") were excluded, because these were nonspecific
and represented an absence of specific targeting to the CEA tumormarker. On CT scan, any noncystic lesion within the liver, nonentericsoft tissue density within the extrahepatic abdomen or pelvis, or abnormal intraluminal thickening of the bowel wall was considered a lesion.Surgically, a discrete area of altered anatomy suggestive or suspiciousof malignancy was defined as a lesion. Pathologically, a lesion wasdefined as histológica! microscopic confirmation of adenocarcinomacompatible with origin from a colorectal primary. Multiple lesions inone organ were each considered pathologically identical if they wereconsidered identical (except for size) by surgical evaluation and at leastone was microscopically evaluated.
Only abdominal lesions with pathological confirmation were utilizedin the statistical lesion analysis. Lesion size (in cm) and lesion location(primary colorectal, hepatic metastasis, or extrahepatic abdominal metastasis/recurrence) were included in the analyses. Pathological classification was used as the absolute status ("gold standard") of all lesions.
Relative to the pathological status, the IS and CT scans (alone andcombined) were scored for each lesion as one of: TP, TN, FP, and FN.
Statistical Methods. The sensitivity, PPV, and accuracy for the twoimaging modalities were calculated using standard statistical formulas(15). To examine the correlation between lesions within a subject, theintraclass correlation coefficient was estimated using the one-way random effect model, in which the dependent variable is a binary variableindicating the degree of agreement between the two imaging modalities(16,17). For these data, the intraclass correlation coefficient was foundto be extremely low and was not statistically significantly different fromzero (using the F test from the one-way analysis of variance). Consequently, multiple lesions from the same subject were treated as independent measurements. McNemar's test of symmetry' was used to
compare the sensitivities and accuracies from the imaging modalities.While the formula for accuracy was used for statistical purposes, thisparameter would be more appropriately considered as "correct classification" of lesion (/.e., benign versus malignant). Statistical significance
was considered as <0.05. All hypothesis tests were two sided unlessotherwise indicated.
On the basis of the study inclusion criteria, both patients and lesionswere selected for the presence of adenocarcinoma. Therefore, the sampling of normal tissue was uncommon. The usual cause for biopsy ofnormal tissues occurred when one test detected a lesion which proved
to be FP, while a second diagnostic test identified no lesion (TN). As aresult, specificity (TN/TN + FP) would have been biased, based uponvery few numbers, and was not included in our analysis.
Both IS and CT were documented to contribute to the diagnosis of primary and metastatic colorectal cancer (Figs. 1-6).Some malignant lesions were identified using both IS and CT(Fig. 1). Other malignant lesions were identified using CT butnot using IS (Fig. 2). As defined above, nonspecific photopenicareas on IS, usually in the liver, were noted and considereduntargeted lesions for purposes of the data analysis. Severalmalignant lesions were identified using IS but not using CT(Fig. 3). Hepatic lesions poorly visualized using planar IS wereoften clearly identified using SPECT (Fig. 4). Some lesionsidentified using IS were localized surgically but did not containCEA-bearing carcinoma pathologically (Fig. 5). Other lesions
were only appreciated after surgical resection and pathologicalevaluation (Fig. 6). Only one lesion identified by surgical exploration to be grossly compatible with carcinoma was biopsiedand not confirmed to be malignant by pathological evaluation.
Sensitivity of IS imaging for colorectal adenocarcinoma was40.5% (49 of 121), of CT was 61.2% (74 of 121) (P = 0.0006),and of IS and CT combined was 72.7% (88 of 121) (Table 1).PPV of IS and CT were similar (83.1 and 88.1%, respectively).Accuracy of CT (61.2%) was higher than IS (44.2%) (P =0.0013).
PPV. For colorectal and hepatic lesions, PPV of both IS andCT imaging was very good (88-100%), but for extrahepaticlesions PPV was somewhat lower (68-80%) (Table 2).
Accuracy. Scan accuracy, a reflection of scan utility for correctly identifying the nature of lesions (benign or malignant),varied with the region. Accuracy of both IS and CT imagingwas high for primary colorectal lesions (85-89%). For hepatic
Fig. 1. A 43-year-old female presented with right lower quadrant pain, constipation, and an abdominal mass. A barium enema showed an apple core lesion inthe ascending colon. Surgery and pathology revealed a 2.5 x 9 x 10-cm firmulcerating tumor mass in the cecum with three foci of palpable enlarged mattedlymph nodes. A, planar posterior scintiscan 72 h following injection of ZCE025(40 mg. 4.9 mCi) demonstrating a focus of intense uptake (arrow) correspondingto the cecal carcinoma. Note the substantial uptake in normal liver superior tothe lesion. B, axial CT scan following i.v. and gastrointestinal contrast demonstrating a large colonie obstructing lesion in the cecal area (arrow) correspondingto the cecal carcinoma.
In the current study, patients who were selected were stagedpresurgically with both '"In anti-CEA antibody (ZCE025) IS
and with conventional CT. Criteria were applied retrospectivelyregarding preoperative testing by IS and CT to make the groupas homogenous as possible. Both IS and CT scans were re-reviewed emphasizing identification of individual lesions in theabdomen. All patients had abdominal exploration with documentation of the individual sites of disease. There were 186lesions identified using IS, CT, or the surgical procedure in the45 patients evaluated. Seven extraabdominal lesions and 32lesions that were surgically documented but that did not havepathological confirmation were excluded from analysis. Thus,this comparison of IS and CT was based upon an analysis of147 surgically explored and pathologically confirmed lesions inthe abdomen using uniform patient and lesion selection criteria.
Primary colorectal cancer lesions were identified correctly ina high percentage of cases by both IS and CT. Almost all ofthese lesions were large (>3.0 cm) and had previously beenidentified by other tests (barium enema, colonoscopy). Theability to detect smaller primary lesions in the colorectum usingIS prior to their identification by conventional modalities remains untested.
previously documented in the nude mouse model that liveraccumulation was due to parenchymal accumulation of a lowmolecular weight break-down product of the '"In-antibody (18,
19). Using size exclusion HPLC analysis of homogenized normal liver, we have demonstrated the same low molecular weightsubstance in humans.4 The use of a novel transition metal
chelate technology recently reported by Hawthorne et al. (20)has led to lower liver accumulation of radioisotope and no lowmolecular weight material on HPLC.5
Current explanations for the photopenic nature of mosthepatic lesions in this study include (a) the high uptake of free"'In and nonspecific '"In-labeled antibody by histologically
4J. D. Beam and B. G. Beatty, unpublished data.5 B. G. Beatty. R. J. Paxton, M. F. Hawthorne, J. D. Beatty, A. Varadarajan,
T. Do, and M. Lewis. Radioimmunodetection using a radiometalcarborane complex (venus flytrap cluster) in an animal model, presented at the Third Conferenceon Radioimmunodetection and Radioimmunotherapy of Cancer, Princeton, NJ,November 1990.
(c) interstitial transport distances and high interstitial pressurein tumors (21 ), and (d) clearance of antigen-antibody complexesof CEA-1 "In monoclonal antibody by the histologically normal
tastases, it was disappointing in its assessment of lymph nodesstatus. Often grossly and histologically malignant nodes werenot localized, while histologically uninvolved nodes drainingknown tumor masses had high '"In content and were thereforereadily visualized. We have observed that these false-positivenodes contained a radiolabeled low molecular weight substancewhen homogenized and run on HPLC." Labeling techniques
that result in less accumulation of radiolabeled low molecularweight catabolites in lymph nodes also may aid in reducing thisproblem. However, the reason for this accumulation in normalnodes is not fully understood. We have found that these lymphnodes contained a higher concentration of CEA than normalbackground tissue, and this was possibly related to the antigen-filtering effect of antigen-processing cells (i.e., macrophages)that reside in lymph nodes (22). These antigen-processing cellsare known to engulf the antigen, degrade the antigen in fragments (8-200 amino acids), and recycle the antigen fragmentsto the cell membrane (23). These membrane-bound fragmentsmay provide the appropriate binding site for '"In anti-CEA
Fig. 3. A 59-year-old male 21 months following resection of cecal carcinomapresented with an increasing plasma CEA (20.7 ng/ml). Surgery and pathologydemonstrated multiple hepatic and extrahepatic foci of metastatic colon adeno-carcinoma. Planar anterior scintiscan 72 h following injection of ZCE025 (40mg, 5.63 mCi) showed one distinct focus within the right abdomen at level L2(arrow) and a less distinct focus in the mid-lower abdomen, overlying approximately L5 (arrow). Axial CT images (not shown) were normal.
B
monoclonal antibody. The above hypothesis is supported byimmunohistological documentation that the CEA in hot lymphnodes is localized in the histiocytes.6
In this report, the focus has been the comparison of the ISand CT modalities for individual lesion identification in contrast to previous studies which concentrated on region, organ,or whole body analysis. The present sensitivity measure wasbased on an individual lesion by lesion analysis. Specifically, ifonly one of 3 lesions was visualized in the liver, the sensitivitywas 33% by lesions analysis, whereas it would be 100% byregion, organ, or whole body analysis. Regional analysis wouldconsider only the presence or absence of disease in the regionand the presence or absence of a positive finding by the IS andCT modalities. Thus, it is less rigorous than the lesion by lesionanalysis, which accounts for the increased sensitivity and accuracy reported using a regional method (9, 10).
Fig. 4. A 70-year-old male was evaluated 17 months following resection ofrectal carcinoma. Surgery and pathology confirmed a 6-cm hepatic adenocarci-noma metastasis. A, planar anterior scintiscan 72 h following injection of ZCE025(40 mg, 6.60 mCi), demonstrating a hot lesion within the liver (arrow). B, coronalSPECT view of the same photophilic liver lesion. C, axial CT view demonstratinga lesion (6x5 cm) in the right lateral lobe of the liver (arrow).
Fig. 5. A 53-year-old female evaluated 19 months following resection of a cecal cancer presented with an elevated CEA (61.5 ng/ml). Surgery and pathologyconfirmed metastatic mucinous adenocarcinoma in the omentum (lesion /1 and in the pelvis (lesion 2). A, planar anterior scintiscan 72 h following injection ofZCE025 (40 mg, 7.12 mCi) demonstrating 2 lesions (arrows). B, coronal SPECT scan demonstrating these same lesions (arrows). C, axial CT view showing theanterior midline periumbilical lesion (/)./), axial CT view of the pelvis showing the lesion at the level of the inferior portion of the iliac bone. (2).
28.2 to 43.7% and from 67.6 to 69.0% for IS and IS+CT, cause this modality was much less specific, reflecting only therespectively.
We excluded photopenic lesions from the positive categorybecause we thought IS was intended to localize specifically totumors with the appropriate tumor marker; thus, it would notbe fair to include lesions that were visualized simply becausethey were nonspecific space-occupying foci seen in relief againsta background of intense uptake (i.e., the liver). On the otherhand, we accepted any abnormal-appearing lesion on CT be-
presence of a mass that was not clearly a benign cyst. Thus, webelieve IS had more potential for identifying the presence andnature of a lesion and we had a higher expectation for itsperformance. This decision is also consistent with our eventualintention of using radiolabeled monoclonal antibodies for ra-dioimmunotherapy, but it does tend to favor CT over IS interms of sensitivity.
Previous reports (9, 10) focused upon the clinical value for5709
to benefit half of a carefully selected population of colorectalcancer patients.
Irrespective of the method of data analysis (lesionai or regional), the basic objective of the IS technique has been the useof a radiolabeled antibody directed against a tumor marker forspecific targeting to tumor bearing the marker. The ability tovisualize the tumor depends on a number of factors includingthe physical characteristics of the radionuclide, the pharmaco-kinetics of the agent, the size and depth of the tumor, and therelative uptake of the radionuclide in tumor and normal background tissue (24). In general, visualization by gamma camerascintigraphy is dependent upon T/NT ratios of isotope uptake.This ratio may be obtained from tissue analysis expressed asunit % ID/kg using a well gamma counter or from apparent
ZCE025 (40 mg, 5.91 mCi) demonstrating one of the hepatic lesions (arrow)which was not visualized on planar images. B, axial CT scan through the domeof the right hepatic lobe, presumably demonstrating the same lesion (arrow). C,axial SPECT view 72 h demonstrating hot portal lymph node (arrow).
contrast values (count density per pixel in the tumor divided bythe count density in adjacent normal tissue) using gammacamera scintigraphy. Minimal T/NT ratios for visualizationappear to be in the range of 1.5-2. For example, of 3 metastaticlesions imaged using phospholipid vesicles, Turner et al. (25)reported an average T/NT ratio of 1.74 with a range of 1.58-2.06. Similarly, the apparent contrast found in four Kaposi's
sarcoma lesions (26) has been reported to be 2.2 when usingthe same agent. At relatively high T/NT ratios smaller, moredeeply seated tumors are visualized, while at lower T/NT ratiosmost tumors are not seen. An exception occurs for lesions withlow uptake embedded in normal tissues of high uptake ofradiolabeled antibody. The resultant very low T/NT ratio leadsto visualization as a region of decreased uptake. However, thegoal of future studies with radiolabeled antibodies will remainto increase uptake in the target tissue and decrease uptake ofthe radionuclide in adjacent background tissue.
The authors gratefully acknowledge the contributions of Merle S.Smith, R.N., (protocol nurse); Russ Kondo (data manager); KathyThomas, R.T., Ron Fomin, R.T., and Joy Bright, C.N.M.T. (nuclearmedicine technologists); and Sarah Farmer Earll, M.A. (secretary).
REFERENCES
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.1. Beatty. J. D., Duda, R. B., Williams, L. E., Sheibani, K., Paxton, R. J.,
Beatty, B. G., Philben, V. J., Werner. J. L., Shively, J. E., Vlahos, W. G.,
surgical, histopathologic and immunohistochemical findings. Radiology, 166:747-752, 1988.Abdcl-Nabi, H. H., Schwartz, A. N., Higano, C. S., Wechter, D. G., andUnger, M. W. Colorectal carcinoma: detection with '"In-anti-carcinoem-bryonic antigen monoclonal antibody ZCE025. Radiology. 164: 617-621,1987.Beatty, J. D., Hyams, D. M., Morton, B. A., Beatty, B. G., Williams, L. E.,Yamauchi, D., Merchant, B., Paxton. R. J., and Shively, J. E. Impact ofradiolabeled antibody imaging on management of colon cancer. Am. J. Surg.,157: 13-19, 1989.Beatty, J. D., Williams, L. E.. Yamauchi, D., Morton, B. A., Beatty, B. G.,Merchant, B., Paxton, R. J., and Shively, J. E. Presurgical imaging withindium labeled anti-CEA for colon cancer staging. Cancer Res. (Suppl.), 50:922s-926s, 1990.Chetanneau, A., Lehur, P. A., Ripoch, D., Peltier, P., Saccavini, J. C.,Vuillez, J. P., Tournemaine, N., Thedrez, P., and Chatal, J. F. Histológica!correlation of 17 prospective immunoscintigraphies of recurrences of colorectal carcinomas using '"In-labeled anti-CEA and(or) 19-9 monoclonalantibodies. Eur. J. Nuclear Med., /5: 302-306, 1989.Molting. T., Schlag, P., and Georgi, P. Current status of immunoscintigraphyin colorectal cancer, results of 5 years' clinical experiences. Eur. J. Surg.Oncol., 4:312-318, 1990.Thompson, W. M., and Halvorsen, R. A., Jr. CT staging of GI malignancies.II. The small bowel, colon and rectum. Invest. Radiol., 22: 96-105, 1987.Meares, C. F., McCall, M. J., Reardan. D. T., Goodwin, D. A., Diamanti,C. I., and McTigue, M. Conjugation of antibodies with bifunctional chelatingagents. Ann. Biochem., 142: 68, 1984.Galen, R. S., and Fink, D. J. Probabilistic approaches to clinical decisionsupport. In: Computer Aids to Clinical Decisions, Vol. 2. Boca Raton, FL:CRC Press, Inc., 1982.Donner, A., and Koval, J. The estimation of intraclass correlation in theanalysis of familial data. Biometrics, 36: 19-25, 1980.Rosner, B. Multivariate methods in ophthalmology with application to otherpaired-data situations. Biometrics, 40: 1025-1035, 1982.Beatty, B. G., O'Connor-Tressel, M., Do, T., Paxton, R. J., and Beatty, J.D. Mechanism of decreasing liver uptake of '"In-labeled anti-CEA monoclo
nal antibody by specific antibody pretreatment in tumor bearing mice. CancerRes. (Suppl.), 50: 846s-851s, 1990.Beatty, J. D., Beatty, B. G., O'Connor-Tressel, M., Do, T., and Paxton, R.
J. Mechanism of tissue uptake and metabolism of radiolabeled antibody: roleof antigen complex formation. Cancer Res. (Suppl.), 50: 840s-845s, 1990.Hawthorne, M. F., Varadarajan, A., Knobler, C. B., Chakrabarti, S., Paxton,R. J., Beatty, B. G., and Curtis, F. L. Radiometallacarboranes as tumorimaging reagents. J. Am. Chem. Soc., 112: 5365-5366, 1990.Jain, R. K. Physiological barriers to delivery of monoclonal antibodies andother macromolecules in tumors. Cancer Res. (Suppl.), 50: 8 14s-8 19s, 1990.Kühn,J. K., Corbisiero, R. M., Buras, R. R., et al. Intraoperative gammadetection probe with presurgical antibody imaging for colon cancer. Arch.Surg., in press, 1991.Vitetta, E. S., Fernandez-Botran, R., Myers, C. D., and Sanders, V. M.Cellular interactions in the humeral immune response. Adv. luminimi., 45:8-19, 1989.Bradwell, A. R., Fairweather, D. S., Dykes, P. W., Keeling, A., Vaughan, A.,and Taylor. J. Limiting factors in the localization of tumors with radiolabeledantibodies. Immunol. Today, 6: 163-169, 1985.Turner, A. F., Presant, C. A., Proffitt, C. A., et al. '"In-labeled liposomes:dosimetry and tumor depiction. Radiology. 166: 761-765, 1988.Presant, C. A., Blayney. D., Profitt, R. T., et al. Preliminary report: imagingof Kaposi's sarcoma and lymphoma in AIDS with '"In-labeled liposomes.Lancet, /: 1307-1309, 1990.
1991;51:5704-5711. Cancer Res Raffael M. Corbisiero, Dave M. Yamauchi, Lawrence E. Williams, et al. AnalysisTomography in Identifying Colorectal Cancer: Individual Lesion Comparison of Immunoscintigraphy and Computerized