Biokinetics of yttrium-90–labeled huBrE-3 monoclonal antibody




This study reports summary biokinetics for 17 patients treated with huBrE-3 antibody labeled with indium-111 (111In) and yttrium-90 (90Y) in a Phase I dose escalation trial.


Patients were infused with huBrE-3 antibody conjugated to 1-p-isothiocyanatobenzyl 3-methyl- and 1-p-isothiocyanatobenzyl 4-methyl-diethylenetriamine pentaacetic acid (MX-DTPA). The huBrE-3 was labeled with increasing amounts of 90Y radioactivity according to the following activity regimen: 10 mCi/m2, 20 mCi/m2, 33 mCi/m2, 50 mCi/m2, and 70 mCi/m2. In addition to the 90Y activity, 3–5 mCi of 111In was labeled to huBrE-3 to serve as an imaging agent. In characterizing the biokinetics of huBrE-3, serial urine and blood samples were acquired. Additionally, whole-body imaging using a scintillation camera was performed at four time points postinfusion.


Cumulative urine data yielded a plot of total-body biologic excretion that was relatively flat. Total body regions of interest derived from nuclear medicine scintigrams decreased according to a monoexponential function with a slope slightly greater than the rate of physical decay. When physical decay was combined with the urine biologic excretion rate, the calculated rate of activity decrease was indistinguishable from the scintigraphic rate of decrease in total-body activity.


The authors concluded from these observations that the radioactivity remains essentially inside the patient, that biologic excretion of activity from the total body is not appreciable, and that the path for biologic excretion of activity that does occur is via the urine. The half-time associated with the β (slow) phase for extraction from the blood averages 40.5 hours. Since large amounts of radioactivity do not appear in the urine, and total-body activity is decreased approximately according to physical decay (64.1 hours), activity must pool elsewhere after leaving the blood. The logical place is the skeleton, with possible selective binding to the bone marrow. Bone marrow biopsies from 4 of 7 patients who consented to serial biopsies were supportive of this conclusion. Cancer 2002;94:1240–8. © 2002 American Cancer Society.

DOI 10.1002/cncr.10292

BrE-3 is an immunoglobulin IgG1 antibody that recognizes an epitope of the human breast epithelial mucin, milk, a protein frequently expressed on the surface of breast carcinoma cells.1 As such, it represents a logical vehicle for the delivery of radioactivity to cancer cells that express this protein. In an effort to decrease the antibody's immunogenicity, a humanized form of the antibody has been created.2, 3 As of the time this article was written, 17 breast cancer patients had been treated with the humanized form of the BrE-3 antibody in a Phase I dose escalation trial at our institution. The antibody was labeled with yttrium-90 (90Y) for therapeutic effect and with a constant amount of indium-111 (111In) as a predictor of 90Y distribution.

Other Phase I radioimmunotherapy trials using a 90Y label have routinely reported dose-limiting toxicity in the 25- to 30-mCi range,4, 5 with the critical organ being the bone marrow. The amounts of 90Y activity intended for infusion in this trial have been sufficient to destroy the bone marrow in all but the first dose cohort. Consequently, this trial has been performed in conjunction with bone marrow stem cell support, allowing us to treat beyond the initial critical organ. This article reports our results with the biokinetics of this antibody in a preliminary group of 17 patients.


Summary of Treatment Plan

Patients were admitted as inpatients to the Clinical Research Center of the University of Colorado Health Sciences Center. In their individual rooms, patients were infused with huBrE-3 antibody conjugated to MX-DTPA. The huBrE-3 was labeled with 90Y radioactivity to achieve a specific activity of approximately 10 mCi/mg. Increasing amounts of radioactive-labeled antibody were infused into patients according to the following activity regimen: 10 mCi/m2, 20 mCi/m2, 33 mCi/m2, 50 mCi/m2, and 70 mCi/m2. Three patients were assigned to each activity cohort, except the last, which was assigned six patients. At the time this article was writted (October 2000), one patient remained to complete the 70-mCi/ m2 activity cohort. In addition to the 90Y activity, 3-5 mCi of 111In was labeled to huBrE-3 to serve as an imaging agent.

To maintain antibody immunoreactivity, the specific activity for the labeling procedure was fixed at 10 mCi 90Y /mg huBrE-3 antibody. The dose cohort and body surface area for each patient consequently determined the amount of labeled antibody protein each patient received. Patients were infused with a total of 50 mg of antibody. The difference between 50 mg and the labeled amount was delivered as unlabeled antibody 1 hour prior to the infusion of labeled antibody. The purpose of this “cold” antibody was to saturate nonspecific binding sites that would otherwise act as a sink for labeled antibody, and to scavenge circulating antigen and mitigate the binding of radioactive carrier molecules to protein not at the cancer cell surface.

To characterize the biokinetics of huBrE-3, serial urine and blood samples were acquired. In addition, whole-body imaging using a scintillation camera was performed at 4 time points postinfusion. Because of the variety of data gathered and the unique methods by which each was processed, a brief description of the data acquisition specifics is given below for each data source. The algorithms used to analyze each data set are also described.

Sample Urine Data Processing

Urine was collected over intervals of 0–12, 12–24, 24–48, 48–72, and 72–96 hours postinfusion. Patients were instructed to urinate into a plastic urine hat placed in the toilet. Nursing staff transferred the urine from the hat to a collection jug fitted with a wide-mouth funnel. No urine was lost during transfer. Total volume for each collection interval was recorded, with 1 cm3 aliquots withdrawn and counted in an automatic gamma counter set for a 90Y energy window (1900–2000 keV). Each sample was counted for 2 minutes, resulting in a count accuracy of less than 0.1% standard deviation. Counts were compared against a reference aliquot drawn from the infusion bag at time t = 0, effectively correcting for physical decay. Counts for each collection interval were scaled up by the collection volume and converted to activity using the reference aliquot.

Each activity was added to previous collection amounts, yielding data representing the cumulative appearance of activity in the urine corrected for physical decay. Subtracting each data point from the infused activity yielded data representing the biologic excretion of activity from the total body via the kidney-and-bladder pathway. This set of data points was fit to a monoexponential function. The rate for biologic elimination via the urine (in hrs−1) was recorded, and its corresponding half-time (in hours) was calculated. In addition, the effective half-time and effective rate for a 90Y label was calculated using the following equations:

equation image

Sample Blood Data Processing

One-mL plasma samples were collected at 1, 4, 8, 12, 24, 48, 72, and 96 hours postinfusion and counted in an automatic gamma counter set for a 90Y energy window (1900–2000 keV). Each sample was counted for 2 minutes, resulting in a count accuracy of less than 0.1% standard deviation. As with the urine data, counts were compared against a reference aliquot drawn from the infusion bag at time t = 0, effectively correcting for physical decay. Counts at each time point were converted to activity using the reference aliquot. The activity of the reference aliquot was derived from proportional volumes, i.e., aliquot activity was assumed related to the infused activity by the ratio (Valiquot/Vinfusion). This assumption implies a well-mixed infusion bag and does not rely on geometry and sensitivity calibration factors to measure absolute activity.

This set of data points was fit to a biexponential function, and the alpha (rapid) and beta (slow) phases for biologic disappearance of activity from the blood recorded. The corresponding half-time (in hours) was subsequently calculated. In addition, the effective rate and effective half-time for a 90Y label was calculated using the same equations detailed in the description for processing the urine data.

Sample Total-Body Scintigram Data Processing

Total-body scans were acquired at approximately 4, 24, 48, and 72 hours postinfusion on a dual-headed Siemens Whole Body Scanner (Hoffman Estates, IL). Both crystals of the scanner were equipped with Siemens medium-energy all-purpose collimators. Images were acquired using dual 15% energy windows centered on the 172 and 247 keV photopeak energies of 111In. Scan speed was set to 5 cm/min, typically requiring 40 minutes to complete the simultaneous anterior and posterior scans. No attempt was made to correct 111In count densities for attenuation and scatter. Count densities were presumed to be predictive of corresponding 90Y activity densities. Data were collected on 14 of the 17 patients. For the remaining three patients (Patients 13, 15, and 16), the 111In tracer failed the quality control requirement of 95% radiochemical purity. These patients consequently did not receive the 111In imaging agent.

Each patient's set of four data points was fit to a monoexponential function, and the rate for disappearance of activity from the total body was recorded in hours−1. The exponential function was forced through the initial infused activity point at time t = 0. The corresponding half-time (in hours) was calculated for comparison to the urine data.

Sample Marrow Biopsy Data Processing

It is especially important to measure (when possible) the magnitude of temporally changing bone marrow activity, because of its radiosensitivity. Seven patients consented to serial bone marrow biopsies. Biopsies were taken from the iliac crest. The time of the initial biopsy and the elapsed time between the first and second biopsy were variable: the first patient was sampled at 24 hours and 168 hours postinfusion, the second and third patients were sampled at 48 and 168 hours postinfusion, and the fourth through seventh patients were sampled at 48 hours and 72 hours postinfusion. Each biopsy was weighed, stored, and counted at the same time as the blood and urine samples. The counting technique changed during the course of the Phase I study: Patients 1, 2, and 3 had their biopsies counted by Cerenkov counting, biopsies from Patients 4 and 5 were counted using a liquid scintillation counter, and biopsies from Patients 6 and 7 were counted using a gamma counter. The two marrow time-activity points were divided by their respective masses and fit to a monoexponential curve. The slope was tabulated for comparison to the rate of physical decay and the rate of total-body excretion.


A plot of total-body activity versus time, as evidenced by the subtraction of decay-corrected activity appearing in the urine, was almost flat for all patients. Figure 1 shows a representative plot. This curve yielded the biologic elimination rate of 90Y from the total body. The plot is an indicator of the minor role played by the kidney-and-bladder pathway in eliminating total-body activity. Table 1 details the total-body effective half-times assuming only urinary excretion. Effective half-times for total-body activity had a mean of 60 hours, with a standard deviation of ± 1.2 hours. One may conclude that biologic excretion via the kidneys was small for all patients.

Figure 1.

Representative time-activity curve for the total body, derived from the subtraction of decay-corrected activity in the urine. (Data derived from a screen capture of the MABDOSE8, 9 dosimetry program.)

Table 1. Total-Body Effective Half-Times, Assuming Only Urinary Excretiona
Dose cohortPatient IndexTotal-body elimination, from cumulative urine counts
λbiol (hr−1)Tbiol (hrs)bλeff w/90Y (hr−1)Teff w/90Y (hrs)c
  • a

    Effective half-times were calculated using the 90Y physical half-life of 64.1 hrs.

  • b

    Mean Tbiol (hrs): 1014.2; standard deviation: 359.8.

  • c

    Mean Teff (hrs): 59.9; standard deviation: 1.2.

No. 1: 10 mCi/m210.000671034.30.0114860.4
No. 2: 20 mCi/m240.00110630.90.0119158.2
No. 3: 33 mCi/m270.000391773.20.0112061.9
No. 4: 50 mCi/m2100.00097716.00.0117858.8
No. 5: 70 mCi/m2130.00086806.10.0116759.4

Total-body activity, as evidenced from the posterior views of whole-body nuclear medicine scintigrams, demonstrated an average effective half-time of 64 hours (Table 2), slightly shorter than the 67-hour half-life of 111In. In order to compare the urine data with the scintigraphic data, one of the data sets needed to be converted: the urine data were counted using a 90Y window, and the scintigraphic data were imaged using a 111In window. Although the two nuclides were similar in decay times (64.1 vs. 67.9 hours), they were not identical. Over the course of a week, this difference increased in significance, with errors compounded by the reality of having to fit sparse data. Consequently, the biologic half-time determined from the urine data was combined with the physical half-life of 111In, and the data in Table 1 were converted for comparison to the scintigraphic imaging data.

Table 2. Total-Body Activity Demonstrated by Posterior-View Total-Body Scintigramsa
Dose cohortPatient indexTotal-body elimination, from scintigrams
λeff (hr−1)Teff (hrs)b
  • a

    Effective half-times should be compared with the 111In physical half-life of 67.9 hrs.

  • b

    Mean Teff (hrs): 64.1; standard deviation: 3.6.

  • c

    Final imaging pt (166 hrs) was not used in fit.

  • d

    The whole-body camera was broken; a series of discrete images were acquired in place of a whole-body image.

No. 1: 10 mCi/m210.0108264.0
No. 2: 20 mCi/m24c0.0106964.8
No. 3: 33 mCi/m270.0110063.0
No. 4: 50 mCi/m2100.0110063.0
No. 5: 70 mCi/m213No imaging data
15No imaging data
16No imaging data

A comparison of total-body effective half-times, as deduced from activity appearance in the urine, with those derived from the whole-body scintigrams demonstrated a negligible difference (Table 3), with the only exceptions being Patients 8 and 9. The observation that these two patients had effective half-times derived from scintigraphic data that were longer than the physical half-life of 111In—a physical impossibility—indicates that these data were probably flawed. Indeed, upon returning to the raw data, it was discovered that the whole-body camera was not operational at the time of these two patients' whole-body scans. Their total-body region-of-interest data had been pieced together from sequences of six LFOV static images, with some overlap between images.

Table 3. Comparison of Total-Body Effective Half-Times from Activity Appearance in Urine with Half-Times Derived from Total-Body Scintigrams
Dose cohortPatient indexTotal-body effective half-time, from cumulative urine countsTotal-body effective half-time, from total-body scintigramsEffective half-time for β (slow) phase of blood activity, from serial blood samples
Teff (hrs)Teff (hrs)Teff (hrs)
No. 1: 10 mCi/m2163.764.036.7
No. 2: 20 mCi/m2461.364.834.3
No. 3: 33 mCi/m2765.463.045.8
No. 4: 50 mCi/m21062.063.038.6
No. 5: 70 mCi/m21362.6No imaging data31.7
1563.3No imaging data40.8
1662.0No imaging data45.4

A plot of the disappearance of activity from the blood was informative. All patients demonstrated a biphasic disappearance pattern (Fig. 2). Table 4 summarizes the data for the alpha (rapid) phase of extraction. The standard deviation was quite large compared with the mean value. In an attempt to explain this variability, effective half-times were plotted as a function of patient body surface area (Fig. 3). Although administered activity was scaled on a body surface area basis, the same total amount of protein was infused into each patient. Larger patients would therefore have a larger body space into which the fixed amount of labeled protein could be extracted. Presumably, larger patients should have a more rapid rate of extraction, characterized by a shorter effective half-time for the alpha (rapid) phase of activity disappearance from the blood. This was not observed. The plot demonstrated no correlation whatsoever between the two variables.

Figure 2.

Representative time-activity curve for the blood, demonstrating a biphasic disappearance. (Data derived from a screen capture of the MABDOSE8, 9 dosimetry program.)

Table 4. Extraction Half-Times for the Alpha (Rapid) Phase of Blood Activitya
Dose cohortPatient indexBody surface area m2α (Rapid) phase of blood activity, from serial blood samples
λbiol (hr−1)Tbiol (hrs)bλeff (hr−1)Teff (hrs)c
  • a

    Effective half-times were calculated using a 90Y physical half-life of 64.1 hrs.

  • b

    Mean Tbiol (hrs): 5.9; standard deviation: 3.9.

  • c

    Mean Teff (hrs): 5.2; standard deviation: 3.1.

No. 1: 10 mCi/m211.940.314602.20.325412.1
No. 2: 20 mCi/m241.80.164344.20.175154.0
No. 3: 33 mCi/m271.660.093757.40.104566.6
No. 4: 50 mCi/m2101.550.305742.30.316552.2
No. 5: 70 mCi/m2131.790.250682.80.261492.7
Figure 3.

Plot of effective half-life for the alpha phase of blood excretion versus patient body surface area. The plot demonstrates no correlation between the two variables.

The effective half-time for the beta phase of activity disappearance from the blood was also variable (Table 5), ranging from 32 to 52 hours and averaging 40 hours. The range of extraction half-times did have one thing in common: they were significantly different from the total-body effective half-times (Table 3). Total-body effective half-times were approximately equal to physical decay half-times. If the rate of activity disappearance from the beta phase of the blood were different, it would imply that the activity was pooling somewhere else in the body.

Table 5. Extraction Half-Times for the Beta (Slow) Phase of Blood Activitya
Dose cohortPatient indexβ (Slow) phase of blood activity, from serial blood samples
λbiol (hr−1)Tbiol (hrs)bλeff (hr−1)Teff (hrs)c
  • a

    Effective half-times were calculated using a 90Y physical half-life of 64.1 hrs.

  • b

    Mean Tbiol (hrs): 120.4; standard deviation: 49.2.

  • c

    Mean Teff (hrs): 40.5; standard deviation: 5.4.

No. 1: 10 mCi/m210.0084082.50.0192236.1
No. 2: 20 mCi/m240.0094673.30.0202734.2
No. 3: 33 mCi/m270.00436159.00.0151745.7
No. 4: 50 mCi/m2100.0071497.10.0179538.6
No. 5: 70 mCi/m2130.0110562.70.0218731.7

The specific activity in cpm/mg for serial bone marrow biopsies counted on a 90Y window increased in four of the seven patients. In the remaining three patients, specific activity decreased. Two of the latter, however, had their second time point sampled at 168 hours (Day 7) and were subjected to Cerenkov counting. Their total counts were quite small and deemed less reliable than the other measurements. The remaining patient's specific activity decreased over a 24-hour period (sample points at 48 and 72 hours) at a rate of 0.0247 hours−1.


This report characterizes the huBrE-3 antibody under the in vivo conditions of escalating 90Y activity for therapeutic affect. The biological half-time for clearance from the total body via the kidneys averages approximately 1000 hours. The distribution about this mean value is quite large, however, having a fractional standard deviation of 35%. The reason this variability makes little difference from a dosimetry standpoint is that the minimum biologic half-time measured—618 hours—is still a factor of 10 longer than the physical half-life of 90Y (approximately 64 hours). Dosimetry calculations are dominated by considerations of physical decay rather than biologic excretion. This point is reiterated in Table 1, with all 17 patients having total-body effective half-times tightly grouped around the mean value of 60 hours. The mean value of 60 hours is comparable to the value of 61 hours reported by Kramer in an 111In imaging study using the same antibody.6

When total-body activity derived from urine data is compared with that derived from posterior whole-body scintigrams, the effective half-times for whole-body elimination are identical in all patients. Stated another way, if one uses the biologic half-time derived from the 90Y data and combines it with the physical half-life for 111In, one obtains the effective half-time measured for the total body from whole-body scintigrams. This means that the decrease in activity is wholly accounted for by physical decay (primary) and urinary excretion (secondary). Postulation of an additional elimination path (such as a fecal route of excretion) is therefore unnecessary.

This observation confirms the tentative conclusion drawn by DeNardo7 that blood pharmacokinetics of 111In-labeled BrE-3 conjugated to MX-DTPA are the same as for the 90Y-labeled version. This conclusion should be extended to encompass claims that urinary excretion is all that is required, aside from physical decay, to account for all of the administered activity. It is unlikely that whole antibody labeled with 111In will interact with the body differently when compared with whole antibody labeled with 90Y. Therefore, the 111In and 90Y catabolism products would be expected to appear in equal abundance. Since the 111In label is more stable than the 90Y, it might be expected to appear in greater abundance in the urine: small catabolism products might be expected to retain their 111In label while losing their 90Y label. The excretion rate via the urine must therefore be at least as great, and possibly greater, than that for 90Y. However, if one adopts for 111In the minimum rate that it can possibly be—the measured rate of 90Y —one wholly accounts for all activity. It then follows that the rate for urinary excretion of 111In must the be same as the rate for urinary excretion of 90Y, and that 111In is an excellent predictor of 90Y when chelated with MX-DTPA to huBrE-3.

In this study, the biologic half-time for 90Y activity in the plasma during the beta extraction phase was comparable to that reported by Kramer et al. for 111In6: 120 hours for this study versus 114 hours in the Kramer et al. study. Again, the correspondence between the two measures serves as additional proof that 111In is an excellent predictor of 90Y activity. On the down side, both sets of measurements have standard deviations of ± 40 hours. This large variability emphasizes that if patient outcomes improve with higher activity amounts, mathematic models must be developed that encapsulate patient-specific kinetics in order for dosimetry to be accurate and predictive.

The inability to correlate body size with the alpha phase of activity elimination from the blood is not wholly troubling. If there were such a relationship, it would imply that the radioactivity was behaving as a freely diffusable tracer. Its absence is indirect evidence that 90Y is bound to large proteins whose extraction from the blood plasma are governed by mechanisms other than simple concentration gradients.

The bone marrow biopsy data from this study are to a certain extent equivocal. Four of the seven patients demonstrated a specific activity increase at the iliac crest over time. This was interesting in that it identified the marrow as a pooling site for radioactivity. Unfortunately, increases in specific activity were not observed in all seven patients who consented to serial biopsies. The possible reasons for this are many but include nonstandardization of the counting procedure, nonuniformity in the time of biopsy, and possible geometric and anatomic differences between bilateral iliac crest sites. The variability that is noted between patients points to a deficiency in our understanding of the underlying variables that determine the distribution of radioactivity as a function of time. The development of nonlinear models that are based on the patient's own anatomy will hopefully resolve these issues and enable the prediction of time-activity curves that are accurate for radioimmunotherapy treatment-planning purposes.


The authors thank the nursing staff of the Clinical Research Center and the nuclear medicine technologists in the Department of Radiology at the University of Colorado Hospital.