Br. J. Surg. 1991, Vol. 78, June, 664670
Int rao perative local izat ion of colorectal cancers using radiolabelled monoclonal anti bod ies Radiation detectors may allow the intraoperative localization of small cancer deposits following administration of radiolabelled tumourassociated antibodies. This technique was evaluated in 16 patients with colorectal tumours (14 cancers, one adenoma, one lipoma) with the In-labelled monoclonal antibody ( M A b ) I C R 2 which recognizes the tumour-associated epithelial membrane antigen ( E M A ) . A t operation counting was carried out (3 x 20 s per site) using a hand-held radiation probe over the primary lesions and any palpable lymph nodes in the mesocolon. The tumour to normal colon (TINC) ratio of counts recorded at operation was more than I.5:I in eight of the 14 patients with cancer (mean(s.d.),154(0.41):I ) and0.91: 1 and 146: I respectively in the two patients with benign tumours. Node to normal colon ratios were l~igherin lymph nodes containing metastases. The uptake of radiolabelled antibody ( T J N C ratio) was higher in EMA-expressing cancers than in those not expressing the target antigen (mean(s.d.), 2.45(@65):I versus 1.40(0.20): I , P = 0 4 1 9 ) . An abdominal tumour model was also developed. Radioactively filled containers of 0.5-10 ml representing tumour deposits were suspended in a tank of ”In solution representing the background activity found in normal tissues. The ratio of radioactivity in the ‘tumour’ to that qf background varied from 2:l to 8: 1. The ‘tumour’ was considered to be detectable if the mean counts recorded over the ‘tumour’ exceeded the mean of counts recorded over background by three standarddeviations. A t a ratio of2: I only ‘tumours’ greater thun 5 ml could be detected with a sodium iodide probe and those over IUml could be detected with a cadmium telluride (CdTe) probe. A t a ratio of 8:1, ‘tumours’ of 0.5 ml could be detected with either probe. A t all ratios artd counting periods the Nu1 probe was more sensitive than the CdTe.
’ ’’
6. R. Davidson, W. A. Waddington, M. D. Short and P. 6. Boulos Department of Surgery, University College and Middlesex School of Medicine and Department of Physics and Bioengineering, University College Hospital, London, UK Correspondence to: Mr 6.R . Davidson, University Department of Surgery, Royal Free
School of Medicine, Rowland Hill Street, London NW3 2PF, UK
Of patients presenting with colorectal cancer less than half will survive 5 years, resulting in over 18000 deaths annually from this disease in the UK’,2. Surgical treatment fails in those undergoing potentially curative resection due to an inability to detect and eradicate local or metastatic tumour at the time of the primary resection. Serial computed tomographic scans and both intraoperative and endoluminal ultrasonography offer significant improvement in staging and in the detection of recurrent colorectal ~ a n c e r ~ -Difficulties ~. arise with these procedures in differentiating tumour from non-tumour tissue'^' and hence interest has turned to the use of radiolabelled tumour-associated monoclonal antibodies as physiological imaging agents. Radioimmunoscintigraphy, however, is rarely able to detect tumour deposits less than 2 c m in diameter9s10.By using an intraoperative y-detecting probe following administration of radiolabelled monoclonal antibodies, smaller lesions may be detected than by external imaging with a y-camera due to the shorter distance between source and detector and a reduction in the proportion of scattered activity detected. This technique was first described in 1984 by Aitken et al. in a clinical study and in an animal xenograft model”. Subsequent animal and human studies, mostly using the monoclonal antibody B72.3
which recognizes the Tag 72 antigen, have confirmed the value of this technique for localizing tumour deposits at operation and some tumour deposits have been localized which would not have otherwise been detected”-”. In the present study a monoclonal antibody has been used which recognizes an epithelial membrane antigen (EMA) expressed by over 80 per cent of colorectal cancers. On immunohistochemical studies this antibody shows greater specificity for colorectal cancer than anticarcinoembryonic an tiger^'^-''. The monoclonal antibody, ICR2, has been labelled with the y-emitting radionuclide I I ‘In whose energy of emissions (171 and 247 keV) and half-life (2.8 days) make it suitable for antibody labelling. This preparation, ‘In-ICR2, has been evaluated clinically for the intraoperative detection of tumour deposits in patients with primary colorectal tumours. As a consequence of this study an experimental model was established to evaluate the potential and limitation of the technique, in particular the minimum size of tumours detectable with variation in the radiation detector employed and the tumour uptake of radioactivity.
’‘
Patients, materials and methods
This work cl’as presented to the Surgical Research Society at Newcastle on 7 Julv 1989.
Clinical study Paticwts. Sixteen patients were studied (six men and ten women) with a median age of 69 years (range 51-83 years). In 14 of these
664
0007 1723!91/060664-07,(
~-
1991 Butterworth Heinemdnn Ltd
Monoclonal antibodies in localization of colorectal cancers: B. R . Davidson et al.
Table 1 Clinical details of the patients ~~
Patient no.
Sex
Diagnosis
Dukes' classification
Site -.
I 2 3 4 5 6 7 8 9 10
II 12
13 14
15 16
60 83 51 71 79 58 67 54 74 76 64 76 13 73 67 56
M F M
Adenocarcinoma Adenocarcinoma Adenocarcinoma
Rectosigmoid Rectosigmoid Hepatic flexure
M F
F
Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma Lipoma Adenocarcinoma Adenoma Adenocarcinoma
Rectum Transverse colon Rectum Sigmoid Sigmoid Sigm oid Sigmoid Sigmoid Splenic flexure Caecum Sigmoid
M M
Adenocarcinoma Adenocarcinoma
Rectosigmoid Sigm oid
F F F F F M F F
Differentiation
.-
B
Moderate Moderate Moderate
C* B B
Moderate Moderate Moderate
C C B B
Moderate Moderate Moderate Moderate
B B
B
Moderate
-
-
C
Well
C*
Moderate Moderate
B
*With liver metastases separated by gel chromatography (Sephadex (325). Radiochemical purity was assessed by thin layer silica gel chromatography. One milligram of antibody was labelled with 100 MBq (2.7 mCi) "'In and was administered intravenously over 15 min having excluded hypersensitivity by skin testing. Following administration, temperature, pulse and blood pressure were monitored half-hourly for 6 h.
Intraoperutive probing. Intraoperative probing was carried out at 3-6 days after 'In-ICR2 administration using a cadmium telluride (CdTe) semiconductor probe designed for the intraoperative detection of " 'In (Radiation Monitoring Devices, Watertown, Massachusetts, USA) (Figure I ) . At the start of the operation and before mobilization of the tumour a minimum of three 20 s counts were recorded with the probe held over three regions of the primary lesion, two sites over normal colon, any lymph nodes palpable within the mesocolon and over both the right and left lobes of the liver. O n completion of probing the operation was carried out as planned. The resected specimen was then probed at the same sites counted 'intraoperatively and for the same period to determine the influence of normal tissue background activity on intraoperative measurements. Biopsies of tumour, taken from the growing edge of an ulcerated lesion (or a representative area if polypoidal), and of normal colon were then weighed and counted in a y-well counter allowing the radiolabelled antibody uptake to be calculated per gram of tissue. Tumour to normal colon (T/NC) ratios for counts obtained at intraoperative and resected specimen probing and the y-well counting of tissue specimens were calculated.
''
Figure 1 The cadmium telluride probe, counting unit and ratemeter consisting of a cadmium telluride detector connected to a counting unit which may be s e t to count over a $xed time period. The rulemeter prorides both u iiisual and audible indication of the number of counts being detected
patients the histology of the resected specimen showed colonic adenocarcinoma (nine Dukes' B and five Dukes' C, two of whom had liver metastases) and in two patients benign tumours were found (one lipoma of the sigmoid colon and one tubulovillous adenoma of the caecum) (Table I ) .
Monoclonal antibo@. The monoclonal antibody ICR2 is an IgG,, antibody which has been raised in the rat by hybridoma technology using extracts of human milk fat globule membrane at the Institute of Cancer Research, Sutton, U K . This antibody recognizes the EMA antigen which is expressed by the majority of adenocarcinomas but especially those of the breast, ovary and colon'g. Before patient use the antibody preparation was screened for the presence of bacteria, viruses and endotoxins (Safepharm, Derby, UK). Antibodj, labelling and administration. T o facilitate radiolabelling with '"In, the monoclonal antibody was first conjugated to the metal chelating agent diethylene triamine pentacetic acid via the bicyclic anhydride (ccDTPA) at a 2:l molar ratio. Any free DTPA was separated by dialysis (Medicell, U K ) against Hepes buffer (pH 7.5)and by gel chromatography on Sephadex G25 (Pharmacia, Upsalla, Sweden). The ICRZ-DTPA conjugate was then labelled with '"In by the method of Carrasquillo et and any unbound ' l l I n was
Br. J. Surg., Vol. 78, No. 6, June 1991
Tumour antigen expression and the uptake of radiolahelleduntibodies. Sections of the resected tumours were stained immunohistochemically using the indirect immunoperoxidase staining method to assess expression of EMA which was graded 0 or 1 4 by the percentage of antigen positive cells23. This system of immunohistochemical grading has been shown previously to be r e p r o d ~ c i b l e ~ ~ . Statistical method.$ in the clinical study'. Results were compared between groups using Student's t test, Spearman rank correlation coefficients and Bland and Altman analysis. E.xperirnenta1 eouluation T o investigate further the potential and limitations of intraoperative radioimmunolocalization a model was designed in which a radioactive tumour lay within a radioactive background, the levels of radioactivity in both corresponding to those found in patient studies. This system has been used to evaluate two currently available intraoperative probes. By investigating a range of tumour volumes and ratios of radioactivity between tumour and background, the minimum tumour size detectable with each of the probes could then be established for a given range of tumour to normal tissue ratios and counting times.
665
Monoclonal antibodies in localization of colorectal cancers: B. I?. Davidson et al.
Table 2 Results
of
the clinical study
TiNC count ratios Patient no.
Diagnosis
Site
Operation
Resected specimen
Well count
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Carcinoma Carcinoma Carcinoma Carcinoma Carcinoma Carcinoma Carcinoma Carcinoma Carcinoma Carcinoma Lipoma Carcinoma Adenoma Carcinoma Carcinoma Carcinoma
Rectosigmoid Rectosigmoid Hepatic flexure Rectum Transverse colon Rectum Sigmoid Sigmoid Sigmoid Sigmoid Sigmoid Splenic flexure Caecum Sigmoid Rectosigmoid Sigmoid
I .66 1.98
4.50 5.1 3 1.23 4.00 2.73 4.45 5.87 6.82 4.22 6.25 3.40 4.90 2.70 3.35 5.15 5.69
3.2 3.4 1.8 2.0 1.2 2.1
N/A 1.64 1.81 1.66 2,13 1.20 0.90 1 .oo
0.9 1 1.32 I .06 2.06 1.57 1.07
N/A 3.5 1.6 2.3 1.6 2.3 1.4 1.9 3.O 2.3
T/NC count ratios, mean counts on tumour to that of normal colon; N/A, not available
Biodislriburion daia. The model was based on data available from studies o n patients with cancer in whom ' "In-labelled monoclonal antibodies had been administered. The percentage of the injected dose of radioactivity taken up by tumour tissue was obtained from the few studies which have provided these data (including the present study) and was taken as 0.001 per cent of the injected dose per gramzs.26. The injected dose of radioactivity of 100 MBq of "'In is an approved dose for patient administration commonly use for localization studies. Tumour to background uptake ratios of from 2: 1 to 8:1 were studied, covering the range reported in clinical s t ~ d i e s ~ ' -The ~ ~ .background activity was calculated from biodistribution data obtained in the present clinical study based on the injected activity, uptake in tumour tissue per gram and the TjNC ratio and tumour to blood ratio found in biopsy The model of /he inira-abdominal tumour. Hollow containers of 0.5-10 ml were used to represent the tumour deposits. The smaller sizes were constructed from test tubes in which the liquid was sealed using dental moulding wax and the larger sizes were formed from latex rubber. These contained a solution of "'In, with the concentration of radioactivity being altered for each tumour to background ratio examined. A tank (30 x 28 x 20 cm) was used to represent the abdominal cavity. This was filled to a depth of 17 cm with a solution of '"In at a concentration of 0.5 kBq/ml to give a volume of 15 litres. Sufficient concentrated HCI was then added to give a final tank pH of less than 2, this being required to prevent aggregation of the I1'In due to the formation of "'In(OH),. The level of radioactivity present in the tank was checked before commencing each probe study by removing an aliquot and counting it in a y-well counter. The count was compared with the counts obtained from an aliquot of the solution used in the preparation of the tumours to ensure that the tumour to background count ratio was kept within 5 per cent of that required. y-detecling probes. Two types of radiation detector were evaluated. The first comprised a sodium iodide (NaI) scintillation detector with a crystal 20 mm thick and 26 mm in diameter connected to a photomultiplier tube, power supply and pulse processing and display modules. The collimation, the lead shielding around the aperture in front of the crystal which dictates the window viewed by the crystal, was designed in house for the intraoperative detection of '"In. This detector system was compared with a commercially available CdTe semiconductor probe'used in the clinical study which again had been customized for use wth '"In and which was supplied with both wide angle (w.a.) and parallel hole (p.h.) collimators (Figure 1 ) .
Tumour positioning. The 'tumours' were positioned in the centre or the tank representing background activity, the upper surface being held 1 cm below the fluid surface. The probe examined was positioned over
666
the 'tumour' with the collimator face parallel to and touching the active background. An energy window for counting of 12@300 keV was chosen for both probes to allow the two y-emissions of '"In (171 and 247 keV) to be counted. Couniing and statisiical analysis. For all 'tumour' sizes and all uptake ratios, counting with both probes was carried out for both 20 and 100 s and repeated three times. The 'tumour' count was obtained by counting directly over and 1 cm above the 'tumour' and the background count was taken at a site remote from the 'tumour', which was chosen to be lOcm laterally. The shorter counting interval was selected as a practical time period for counting over one site at operation. Because the technique of tumour detection by intraoperative probing of radiolabelled monoclonal antibodies is based on detecting a difference between tumour and background counts of radioactivity, a longer counting period was also investigated to allow the importance of higher counts and hence improved counting statistics to be analysed. The statistical significance of the difference in counts was assessed by calculating the number of standard deviations of the difference between the mean count obtained over the tumour and that over background (decay corrected) using Poisson s t a t i s t i ~ s. ~Three ~ ~ ~ standard ~.~~ deviations of the difference between the mean tumour and background counts was taken as a significant level of detectability as this has less than a 1 per cent chance of being due to random counting errors.
Results Clinical study (Table 2 ) The T/NC ratio of counts was greater than 1 5 : l in eight of the 14 patients with cancer (mean(s.d.), 1.54(0.41):1). The two benign tumours showed no selective uptake with ratios of0.91:1 and 1.06: 1 respectively. The correlation between the counting ratio found at operation and those found on y-well counting of biopsies, taken as the absolute uptake of "'In-ICR2, was poor (Spearman rank correlation coefficient n = 14, r=0.19, P = 0 5 2 ) (Figure 2). Probing over the resected specimen resulted in higher T/NC ratios than those found at operation (mean(s.d.), 4.61(1.25):1 versus 1.46(0.43):1, P
Int rao perative local izat ion of colorectal cancers using radiolabelled monoclonal anti bod ies Radiation detectors may allow the intraoperative localization of small cancer deposits following administration of radiolabelled tumourassociated antibodies. This technique was evaluated in 16 patients with colorectal tumours (14 cancers, one adenoma, one lipoma) with the In-labelled monoclonal antibody ( M A b ) I C R 2 which recognizes the tumour-associated epithelial membrane antigen ( E M A ) . A t operation counting was carried out (3 x 20 s per site) using a hand-held radiation probe over the primary lesions and any palpable lymph nodes in the mesocolon. The tumour to normal colon (TINC) ratio of counts recorded at operation was more than I.5:I in eight of the 14 patients with cancer (mean(s.d.),154(0.41):I ) and0.91: 1 and 146: I respectively in the two patients with benign tumours. Node to normal colon ratios were l~igherin lymph nodes containing metastases. The uptake of radiolabelled antibody ( T J N C ratio) was higher in EMA-expressing cancers than in those not expressing the target antigen (mean(s.d.), 2.45(@65):I versus 1.40(0.20): I , P = 0 4 1 9 ) . An abdominal tumour model was also developed. Radioactively filled containers of 0.5-10 ml representing tumour deposits were suspended in a tank of ”In solution representing the background activity found in normal tissues. The ratio of radioactivity in the ‘tumour’ to that qf background varied from 2:l to 8: 1. The ‘tumour’ was considered to be detectable if the mean counts recorded over the ‘tumour’ exceeded the mean of counts recorded over background by three standarddeviations. A t a ratio of2: I only ‘tumours’ greater thun 5 ml could be detected with a sodium iodide probe and those over IUml could be detected with a cadmium telluride (CdTe) probe. A t a ratio of 8:1, ‘tumours’ of 0.5 ml could be detected with either probe. A t all ratios artd counting periods the Nu1 probe was more sensitive than the CdTe.
’ ’’
6. R. Davidson, W. A. Waddington, M. D. Short and P. 6. Boulos Department of Surgery, University College and Middlesex School of Medicine and Department of Physics and Bioengineering, University College Hospital, London, UK Correspondence to: Mr 6.R . Davidson, University Department of Surgery, Royal Free
School of Medicine, Rowland Hill Street, London NW3 2PF, UK
Of patients presenting with colorectal cancer less than half will survive 5 years, resulting in over 18000 deaths annually from this disease in the UK’,2. Surgical treatment fails in those undergoing potentially curative resection due to an inability to detect and eradicate local or metastatic tumour at the time of the primary resection. Serial computed tomographic scans and both intraoperative and endoluminal ultrasonography offer significant improvement in staging and in the detection of recurrent colorectal ~ a n c e r ~ -Difficulties ~. arise with these procedures in differentiating tumour from non-tumour tissue'^' and hence interest has turned to the use of radiolabelled tumour-associated monoclonal antibodies as physiological imaging agents. Radioimmunoscintigraphy, however, is rarely able to detect tumour deposits less than 2 c m in diameter9s10.By using an intraoperative y-detecting probe following administration of radiolabelled monoclonal antibodies, smaller lesions may be detected than by external imaging with a y-camera due to the shorter distance between source and detector and a reduction in the proportion of scattered activity detected. This technique was first described in 1984 by Aitken et al. in a clinical study and in an animal xenograft model”. Subsequent animal and human studies, mostly using the monoclonal antibody B72.3
which recognizes the Tag 72 antigen, have confirmed the value of this technique for localizing tumour deposits at operation and some tumour deposits have been localized which would not have otherwise been detected”-”. In the present study a monoclonal antibody has been used which recognizes an epithelial membrane antigen (EMA) expressed by over 80 per cent of colorectal cancers. On immunohistochemical studies this antibody shows greater specificity for colorectal cancer than anticarcinoembryonic an tiger^'^-''. The monoclonal antibody, ICR2, has been labelled with the y-emitting radionuclide I I ‘In whose energy of emissions (171 and 247 keV) and half-life (2.8 days) make it suitable for antibody labelling. This preparation, ‘In-ICR2, has been evaluated clinically for the intraoperative detection of tumour deposits in patients with primary colorectal tumours. As a consequence of this study an experimental model was established to evaluate the potential and limitation of the technique, in particular the minimum size of tumours detectable with variation in the radiation detector employed and the tumour uptake of radioactivity.
’‘
Patients, materials and methods
This work cl’as presented to the Surgical Research Society at Newcastle on 7 Julv 1989.
Clinical study Paticwts. Sixteen patients were studied (six men and ten women) with a median age of 69 years (range 51-83 years). In 14 of these
664
0007 1723!91/060664-07,(
~-
1991 Butterworth Heinemdnn Ltd
Monoclonal antibodies in localization of colorectal cancers: B. R . Davidson et al.
Table 1 Clinical details of the patients ~~
Patient no.
Sex
Diagnosis
Dukes' classification
Site -.
I 2 3 4 5 6 7 8 9 10
II 12
13 14
15 16
60 83 51 71 79 58 67 54 74 76 64 76 13 73 67 56
M F M
Adenocarcinoma Adenocarcinoma Adenocarcinoma
Rectosigmoid Rectosigmoid Hepatic flexure
M F
F
Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma Lipoma Adenocarcinoma Adenoma Adenocarcinoma
Rectum Transverse colon Rectum Sigmoid Sigmoid Sigm oid Sigmoid Sigmoid Splenic flexure Caecum Sigmoid
M M
Adenocarcinoma Adenocarcinoma
Rectosigmoid Sigm oid
F F F F F M F F
Differentiation
.-
B
Moderate Moderate Moderate
C* B B
Moderate Moderate Moderate
C C B B
Moderate Moderate Moderate Moderate
B B
B
Moderate
-
-
C
Well
C*
Moderate Moderate
B
*With liver metastases separated by gel chromatography (Sephadex (325). Radiochemical purity was assessed by thin layer silica gel chromatography. One milligram of antibody was labelled with 100 MBq (2.7 mCi) "'In and was administered intravenously over 15 min having excluded hypersensitivity by skin testing. Following administration, temperature, pulse and blood pressure were monitored half-hourly for 6 h.
Intraoperutive probing. Intraoperative probing was carried out at 3-6 days after 'In-ICR2 administration using a cadmium telluride (CdTe) semiconductor probe designed for the intraoperative detection of " 'In (Radiation Monitoring Devices, Watertown, Massachusetts, USA) (Figure I ) . At the start of the operation and before mobilization of the tumour a minimum of three 20 s counts were recorded with the probe held over three regions of the primary lesion, two sites over normal colon, any lymph nodes palpable within the mesocolon and over both the right and left lobes of the liver. O n completion of probing the operation was carried out as planned. The resected specimen was then probed at the same sites counted 'intraoperatively and for the same period to determine the influence of normal tissue background activity on intraoperative measurements. Biopsies of tumour, taken from the growing edge of an ulcerated lesion (or a representative area if polypoidal), and of normal colon were then weighed and counted in a y-well counter allowing the radiolabelled antibody uptake to be calculated per gram of tissue. Tumour to normal colon (T/NC) ratios for counts obtained at intraoperative and resected specimen probing and the y-well counting of tissue specimens were calculated.
''
Figure 1 The cadmium telluride probe, counting unit and ratemeter consisting of a cadmium telluride detector connected to a counting unit which may be s e t to count over a $xed time period. The rulemeter prorides both u iiisual and audible indication of the number of counts being detected
patients the histology of the resected specimen showed colonic adenocarcinoma (nine Dukes' B and five Dukes' C, two of whom had liver metastases) and in two patients benign tumours were found (one lipoma of the sigmoid colon and one tubulovillous adenoma of the caecum) (Table I ) .
Monoclonal antibo@. The monoclonal antibody ICR2 is an IgG,, antibody which has been raised in the rat by hybridoma technology using extracts of human milk fat globule membrane at the Institute of Cancer Research, Sutton, U K . This antibody recognizes the EMA antigen which is expressed by the majority of adenocarcinomas but especially those of the breast, ovary and colon'g. Before patient use the antibody preparation was screened for the presence of bacteria, viruses and endotoxins (Safepharm, Derby, UK). Antibodj, labelling and administration. T o facilitate radiolabelling with '"In, the monoclonal antibody was first conjugated to the metal chelating agent diethylene triamine pentacetic acid via the bicyclic anhydride (ccDTPA) at a 2:l molar ratio. Any free DTPA was separated by dialysis (Medicell, U K ) against Hepes buffer (pH 7.5)and by gel chromatography on Sephadex G25 (Pharmacia, Upsalla, Sweden). The ICRZ-DTPA conjugate was then labelled with '"In by the method of Carrasquillo et and any unbound ' l l I n was
Br. J. Surg., Vol. 78, No. 6, June 1991
Tumour antigen expression and the uptake of radiolahelleduntibodies. Sections of the resected tumours were stained immunohistochemically using the indirect immunoperoxidase staining method to assess expression of EMA which was graded 0 or 1 4 by the percentage of antigen positive cells23. This system of immunohistochemical grading has been shown previously to be r e p r o d ~ c i b l e ~ ~ . Statistical method.$ in the clinical study'. Results were compared between groups using Student's t test, Spearman rank correlation coefficients and Bland and Altman analysis. E.xperirnenta1 eouluation T o investigate further the potential and limitations of intraoperative radioimmunolocalization a model was designed in which a radioactive tumour lay within a radioactive background, the levels of radioactivity in both corresponding to those found in patient studies. This system has been used to evaluate two currently available intraoperative probes. By investigating a range of tumour volumes and ratios of radioactivity between tumour and background, the minimum tumour size detectable with each of the probes could then be established for a given range of tumour to normal tissue ratios and counting times.
665
Monoclonal antibodies in localization of colorectal cancers: B. I?. Davidson et al.
Table 2 Results
of
the clinical study
TiNC count ratios Patient no.
Diagnosis
Site
Operation
Resected specimen
Well count
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Carcinoma Carcinoma Carcinoma Carcinoma Carcinoma Carcinoma Carcinoma Carcinoma Carcinoma Carcinoma Lipoma Carcinoma Adenoma Carcinoma Carcinoma Carcinoma
Rectosigmoid Rectosigmoid Hepatic flexure Rectum Transverse colon Rectum Sigmoid Sigmoid Sigmoid Sigmoid Sigmoid Splenic flexure Caecum Sigmoid Rectosigmoid Sigmoid
I .66 1.98
4.50 5.1 3 1.23 4.00 2.73 4.45 5.87 6.82 4.22 6.25 3.40 4.90 2.70 3.35 5.15 5.69
3.2 3.4 1.8 2.0 1.2 2.1
N/A 1.64 1.81 1.66 2,13 1.20 0.90 1 .oo
0.9 1 1.32 I .06 2.06 1.57 1.07
N/A 3.5 1.6 2.3 1.6 2.3 1.4 1.9 3.O 2.3
T/NC count ratios, mean counts on tumour to that of normal colon; N/A, not available
Biodislriburion daia. The model was based on data available from studies o n patients with cancer in whom ' "In-labelled monoclonal antibodies had been administered. The percentage of the injected dose of radioactivity taken up by tumour tissue was obtained from the few studies which have provided these data (including the present study) and was taken as 0.001 per cent of the injected dose per gramzs.26. The injected dose of radioactivity of 100 MBq of "'In is an approved dose for patient administration commonly use for localization studies. Tumour to background uptake ratios of from 2: 1 to 8:1 were studied, covering the range reported in clinical s t ~ d i e s ~ ' -The ~ ~ .background activity was calculated from biodistribution data obtained in the present clinical study based on the injected activity, uptake in tumour tissue per gram and the TjNC ratio and tumour to blood ratio found in biopsy The model of /he inira-abdominal tumour. Hollow containers of 0.5-10 ml were used to represent the tumour deposits. The smaller sizes were constructed from test tubes in which the liquid was sealed using dental moulding wax and the larger sizes were formed from latex rubber. These contained a solution of "'In, with the concentration of radioactivity being altered for each tumour to background ratio examined. A tank (30 x 28 x 20 cm) was used to represent the abdominal cavity. This was filled to a depth of 17 cm with a solution of '"In at a concentration of 0.5 kBq/ml to give a volume of 15 litres. Sufficient concentrated HCI was then added to give a final tank pH of less than 2, this being required to prevent aggregation of the I1'In due to the formation of "'In(OH),. The level of radioactivity present in the tank was checked before commencing each probe study by removing an aliquot and counting it in a y-well counter. The count was compared with the counts obtained from an aliquot of the solution used in the preparation of the tumours to ensure that the tumour to background count ratio was kept within 5 per cent of that required. y-detecling probes. Two types of radiation detector were evaluated. The first comprised a sodium iodide (NaI) scintillation detector with a crystal 20 mm thick and 26 mm in diameter connected to a photomultiplier tube, power supply and pulse processing and display modules. The collimation, the lead shielding around the aperture in front of the crystal which dictates the window viewed by the crystal, was designed in house for the intraoperative detection of '"In. This detector system was compared with a commercially available CdTe semiconductor probe'used in the clinical study which again had been customized for use wth '"In and which was supplied with both wide angle (w.a.) and parallel hole (p.h.) collimators (Figure 1 ) .
Tumour positioning. The 'tumours' were positioned in the centre or the tank representing background activity, the upper surface being held 1 cm below the fluid surface. The probe examined was positioned over
666
the 'tumour' with the collimator face parallel to and touching the active background. An energy window for counting of 12@300 keV was chosen for both probes to allow the two y-emissions of '"In (171 and 247 keV) to be counted. Couniing and statisiical analysis. For all 'tumour' sizes and all uptake ratios, counting with both probes was carried out for both 20 and 100 s and repeated three times. The 'tumour' count was obtained by counting directly over and 1 cm above the 'tumour' and the background count was taken at a site remote from the 'tumour', which was chosen to be lOcm laterally. The shorter counting interval was selected as a practical time period for counting over one site at operation. Because the technique of tumour detection by intraoperative probing of radiolabelled monoclonal antibodies is based on detecting a difference between tumour and background counts of radioactivity, a longer counting period was also investigated to allow the importance of higher counts and hence improved counting statistics to be analysed. The statistical significance of the difference in counts was assessed by calculating the number of standard deviations of the difference between the mean count obtained over the tumour and that over background (decay corrected) using Poisson s t a t i s t i ~ s. ~Three ~ ~ ~ standard ~.~~ deviations of the difference between the mean tumour and background counts was taken as a significant level of detectability as this has less than a 1 per cent chance of being due to random counting errors.
Results Clinical study (Table 2 ) The T/NC ratio of counts was greater than 1 5 : l in eight of the 14 patients with cancer (mean(s.d.), 1.54(0.41):1). The two benign tumours showed no selective uptake with ratios of0.91:1 and 1.06: 1 respectively. The correlation between the counting ratio found at operation and those found on y-well counting of biopsies, taken as the absolute uptake of "'In-ICR2, was poor (Spearman rank correlation coefficient n = 14, r=0.19, P = 0 5 2 ) (Figure 2). Probing over the resected specimen resulted in higher T/NC ratios than those found at operation (mean(s.d.), 4.61(1.25):1 versus 1.46(0.43):1, P
