Eur J Nucl Med (1988) 14:498-503 European N u c l e a r Journal of

Medicine © Springer-Verlag 1988

SPECT using Bremsstrahlung to quantify 9Oy uptake in Baker's cysts: Its application in radiation synovectomy of the knee T. Smith 1, J.C.W. Crawley 1, D.J. Shawe 2, and J.M. Gumpel 2 1 Radioisotopes Division, Clinical Research Centre and 2 Rheumatology Department, Northwick Park Hospital, Harrow, Middlesex, UK

Abstract. The use of SPECT with Bremsstrahlung radiation has been investigated in studies on patients undergoing 9Oy therapy for persistent synovitis of the knee. In particular, its value in the estimation of 9oy uptake into Baker's cysts was assessed and, to this end, realistic 'knee phantoms' were employed in order to calibrate for cysts of different size. Problems associated with the measurement of the ex- tensive Bremsstrahlung spectrum and the estimation of cyst volume have been discussed. It is shown that, although the apparent volume of a cyst is markedly dependent on the chosen count rate threshold, volumes greater than about 30 ml can be estimated with reasonable accuracy using a threshold of 50%. The uptake of 9oy in cysts, measured on 3 occasions within the first 2 days in 10 patients, showed wide variation (0%-40%) between patients and was poorly related to the size of cysts on arthrograms and to the clinical response to therapy. In these studies, the ability to analyse SPECT slices provided a distinct advantage over planar imaging for discriminating between 9oy uptake in cysts and adjacent sites. Retention of 9oy in the total knee was also widely variable, with losses of 2%-38% observed 2 days after injection which, in general, were not fully accounted for by uptake in liver or lymph nodes. The changing distri- bution of 9oy colloid in the knee during the first two days, as observed in some patients, might explain part of the discrepancy.

Key words: SPECT - 9oy _ Bremsstrahlung - Radiation synoveetomy - Persistent synovitis - Baker's cyst

The use of fl emitting radionuclides has been widely applied in the radiation treatment of various pathological condi- tions where the treatment region is only a few mm thick. This technique is optimal, for example, for treatment of cystic (Kobayashi et al. 1981) or intracavitary (Croll and Brady 1979) lesions where, in general, fl emitters of appro- priate energy are selected to administer an intense localized dose to target membranes with minimal dose delivered else- where. For some purposes these conditions are best achieved by the use of pure fl emitters, so avoiding the effects of unnecessary gamma radiation. Thus, a2p (Emax: 1.71 MeV) has been used to treat cranial cystic neoplasms (Balachandran et al. 1985; McGuire et al. 1986), for intra- peritoneal therapy of ovarian and endometrial cancer

Offprint requests to: T. Smith

(Pezner et al. 1978; Sullivan et al. 1983) and for treatment of malignant pleural effusions (Croll and Brady 1979) and other pathological disorders (Balachandran et al. 1985), whilst 9oy (Emax: 2.3 MeV) has been used in intraarticular therapy to perform radiation synovectomy in skeletal joints (Gumpel et al. 1973; Spooren et al. 1985). Imaging the dis- tribution of radioactivity is an important adjunct of such treatment, both to verify the accuracy of targeting the ad- ministered activity and to examine the fine structure of the treatment field (Kaplan et al. 1981), since these have bear- ing on the therapeutic response. However, imaging Brems- strahlung radiation from energetic pure fl emitters presents particular problems (Crawley and Smith 1986): the exten- sive continuous spectrum cannot be ideally matched to gamma camera collimators and this usually necessitates a compromise between counting statistics and spectral range, limiting the image quality that can be achieved.

For a complete assessment of the variation of activity distribution within a treatment region, some form of tomog- raphy is desirable. Boye et al. (1984) used a tomographic scanner to examine the distribution of 32p in the peritone- um as an aid in the administration procedure and dosi- merry. We have explored the value of using single photon emission computed tomography (SPECT) to image the 9oy distribution in the knees of patients undergoing treatment for persistent synovitis and in whom there was clinical evi- dence of a Baker's cyst (Baker 1877). These cysts, which are of variable size, interconnect with the synovial space and thus have the potential to take up some of the 9Oy injected into the synovial capsule. The object of this study was to attempt to quantify the uptake of 9oy in Baker's cysts in order to assess its implications with respect to the therapeutic response of the treated synovium. Relationships between cyst size, uptake of 9oy and therapeutic effect ob- served at follow up, if found, might suggest modification of the treatment regimen on an individual basis, and the paper describes our experience using SPECT of Bremsstrah- lung to examine such possibilities.

Materials and methods

Ten patients (5 male, 5 female) aged 54 to 76 years were studied. All had received 90y for treatment of persistent synovitis of the knee and all showed clinical evidence of a Baker's cyst which was confirmed in nine patients by standard arthrography (Taylor 1969). Injection doses of 185 MBq (5 mCi) 9°y-silicate (Amersham International) were checked using a calibrated dosemeter (Capintec). Fol-

499

L0 im

C

(l)

~-- Window

5. I I I

100 200 300

keY Fig. 1. Bremsstrahlung spectrum of 9oy obtained with the gamma camera fitted with a medium energy collimator, showing the se- lected energy range (60-200 keV)

lowing injection into the synovial cavity, the knee was ban- daged for 48 h to prevent flexure. These studies were carried out with the approval of the Northwick Park Hospital Ethi- cal Committee.

SPECT measurements were made using an IGE (Inter- national General Electric Co., New York) 400AT rotating gamma camera, fitted with a medium energy collimator, to detect 9oy Bremsstrahlung. A wide energy window (60-200 keV) was chosen to include most of the continuous spectrum (Fig. 1). Each tomographic measurement involved the acquisition of 64 planar views of the knee at equal an- gles within a complete rotation of the camera. The selected time for each view (15 to 25 s) was determined from the observed count rate, and the total number of counts ob- tained in a complete measurement was about 5 x 105. The radius of rotation was made as small as practicable and the same radius was used for all measurements on patients and calibration phantoms. In 9 patients, measurements were made on 3 occasions at approximately 1-2, 24 and 46 h postinjection: 1 patient was measured at 2 and 46 h only. All counts were corrected for radioactive decay of 907.

Following correction of all images for nonuniformity of the camera, tomographic slices (0.63 cm) of the treated knee were reconstructed by filtered backprojection using a STAR computer and IGE software. A Ramp-Hanning filter with a threshold of 0.5 pixel/cycle was chosen for the reconstruction and no correction was made for attenuation. The size and shape of a Baker's cyst were usually best estab- lished from sagittal slices and consequently these were used for quantification of uptake, but transaxial and coronal slices were also analysed where necessary. Estimates were made of the total 907 activity in the knee and of the propor- tion of injected 907 taken up in the Baker's cyst for each measurement. These were calculated respectively from the integrated counts (counts/s per MBq 907) in reconstructed slices which exceeded a threshold value set at 10% of the maximum pixel count within the family of slices, consider- ing firstly the total knee and secondly the cyst alone. In addition, the volume of distribution of 9oy in the cyst was estimated using a computer program which integrated the number of voxels (cubic pixels) of volume 0.25 cm 3 enclosed within a 50% count rate threshold, for the cyst region alone.

Phantoms were used to calibrate the uptake of 9oy in cysts, and were measured under conditions identical to those employed for patients. The phantoms consisted of human post mortem femur and tibia bones supported on a perspex frame placed inside water filled perspex cylinders

of diameter 10.0 and 12.5 cm. Studies were carried out using sources of various sizes (4-122 ml) containing known activi- ties of 9oy and positioned behind the 'knee ' joint. Further measurements were made with the same sources positioned on the central axis of the two water filled cylinders after removal of the bones. In all measurements on phantoms, a water filled polyethylene cylinder was used to simulate the scattering and attenuation due to the untreated leg. The phantoms were used initially to determine the variation of counting efficiency with threshold level (% of maximum pixel count) for the range of sources. A relatively low threshold value (10%) was then applied to provide calibra- tion factors relating SPECT counting efficiency (counts. s- 1 MBq- 1) to source volume. For the present purpose, count. s- 1 is calculated as the total counts divided by the imaging time for a single view. In addition, the apparent volume at different count rate thresholds was determined for each source, from which the appropriate threshold level required to give an accurate estimate of the volume of each source was found. The calibration factor appropriate for a particu- lar patient's knee was interpolated from the results of the phantom measurements after estimating the dimensions of the knee by direct measurement or from radiographs. The equivalent diameter of the knee was calculated as the mean of the anterior and lateral widths and this value fell within the range of the phantoms (10.0-12.5 cm) for all 10 pa- tients.

Planar anterior views of the liver at 46 h were used in 8 patients to monitor any leakage of 9oy from the knee, a suitable phantom being used to estimate liver uptake. A similar estimate of lymph node activity was made in one patient showing uptake in the groin.

Results

The results of measurements on phantoms are summarized in Figs. 2a 2d. Fig. 2a shows the data obtained with the 10.0 cm diameter phantom containing post mortem bones, where counting efficiency was measured as a function of the count rate threshold for 6 9oy sources of differing vol- ume: the variation shown was typical of that observed with the other phantoms. Fig. 2b shows calibration curves relat- ing counting efficiency (at 10% threshold) to source vol- ume, and includes all data obtained with both the 10.0 cm and 12.5 cm diameter phantoms with and without bones. Fig. 2b was used to determine the 9Oy activity in a Baker's cyst after estimates of the equivalent diameter of the knee and the volume of the cyst had been made. Fig. 2c illus- trates, for the 10.0 cm diameter phantom with included bones, the dependence of the estimated volume of a given source on the threshold value employed, and Fig. 2 d shows that the threshold required to give a correct estimate of the true volume approximates to a value of 50% for vol- umes greater than about 30 ml. Consequently, a 50% threshold was used to estimate the volume of cysts which included 9oy, in order to obtain calibration factors from Fig. 2b.

Retention of 9Oy in the total knee, using a 10% thresh- old, is shown in Fig. 3 for the 10 patients. The mean initial count rate was 213404-_2175 counts, s-1, giving a normal- ised value of 115+l l .8coun t s . s -1. MBq - t (4260_+ 435 counts, s- 1 -mCi- 1) 9oy, which is consistent with the phantom data of Fig. 2b for the larger source volumes.

500

1.6-

1.4- o

1.2-

"7 t - -

0.8- %

"~ 0.6- g

0.4-

0.2-

D •

!o o

;0urce Volume {ml)

• 122 a 62 • 35 z, 17

= A I0

o 4

• o •

o

-6

-5

-4

'7

3 E

2 ~ o~

1

> - o ' 7

1.4-

1.2-

l -

b,

c~ 0.8-

0.6

O

/

x' Phantom dia. (cm) Bones 10.0 12.5

+ O zx

i 0 20 30 40 50 60 0 80 0 20 40 60 80 100 120

a Threshold fro) b Volume ofg°Y source (ml)

300-

E

L. g

> - o

E

o

E

2,.

200-

100-

-5

-4.5

,.-t

- 4

'T -3.5 L5

E

-3 c

-2.5

2

I 140

0 I I I I I I I 0 10 20 30 40 50 60 70 80 20 30 40 50 60 70 80

C Threshold I%) d Threshold for correct estimation of source volume 1%)

120-

100-

g 80- v l > -

,~ 60-

40-

20-

0 m 0 10

Fig. 2a-d. Data obtained using 6 9oy sources of different volume (4--122 ml) in 'knee' phantoms: a relationship between counting efficiency and threshold value for the 10.0 cm diameter phantom, b calibration curves relating counting efficiency to source volume for all phantoms with (+ ) or without ( - ) included bones, e variation of apparent source volume with threshold value for the 10.0 cm diameter phantom (source volumes indicated beside each curve), d relationship between source volume and the threshold required to estimate true volume

There was a variable decrease in knee retent ion with time, the largest change occurring within the first 24 h in most patients. In 2 cases retent ion remained reasonably constant over 46 h and in 1 there was a l inear fall. In the 10 patients, the mean apparen t fall in knee retent ion over 46 h was 15.5% __ 12.2% (SD). Low values of liver uptake were observed in all eight patients who underwent liver counting and in one pat ient there was measurable uptake in the groin, presumably due to accumulat ion of 90y in lymph nodes. In the lat ter patient , the deficit at 46 h in the knee (6%) was, to a fair approximat ion , accounted for by uptake in liver (4%) and groin (2%). However, all other liver uptakes were est imated to be about 3% or less and in several cases were much lower than the observed fall in knee retention.

Serial estimates of the 9oy content o f Baker 's cysts in ten patients, on three occasions during the first two days following injection, are given in Table 1. These activities are also presented in terms of the percentage of the admin-

istered activity and similar values (in parentheses) were ob- tained when the count rate from the cyst was expressed as a percentage of the initial count rate from the total knee. The est imated uptake o f 9oy in cysts ranged from 0% to about 40% of the injected activity, with a mean value of 9.4%. One value (40%) was significantly higher than all other uptakes; which did not exceed 20%, and had a mean value of 6.3%. The selected sagittal, t ransaxial and coronal tomographic slices shown in Fig. 4 were obtained for pa- tient 2, who had the highest measured uptake of 9oy in a Baker 's cyst o f only modera te size, as determined from ar thrography.

Discuss ion

The Bremsstrahlung spectrum of 9oy (Fig. 1) obtained us- ing a medium energy, paral lel hole col l imator illustrates the problem of choice of col l imator and spectral range. On account of the low efficiency of Bremsstrahlung produc-

501

IO0 -

9O

8O o

= 70

6 0 - ¢ - ,_00

5 0 - C~

4 0 - 1 I I 0 24 46

Time (hours)

Fig. 3. Retention of 9Oy in the total knee during the first 2 days after injection

tion it was necessary to choose a wide energy window (60-200 keV) a l though the low intensity high energy range was avoided to reduce image degradat ion due to septal pen- etration. Even so, resolut ion of the system was poor ; a line source of 9oy in a 20 cm water phan tom having a F W H M of about 3 cm (Crawley and Smith 1986). The choice of appropr ia te col l imator is not an obvious one. F o r example, using a similar gamma camera and computer system to ours, Balachandran et al. (1985) used a low ener- gy, paral lel hole col l imator and a 40-150 keV window to image the Bremsstrahlung radia t ion from 32p, while Boye et al. (1984) used a medium energy col l imator with a Sie- mens camera and a 40-180 keV window for the same pur- pose. We did not explore the use of al ternative energy win- dows and coll imators, but any improvement achieved by doing so is likely to be minimal.

At tenua t ion problems associated with the Bremsstrah- lung spectrum from 9oy within such an inhomogeneous tissue region as the knee are also likely to be formidable and we have encountered difficulties in determining the ap- propr ia te a t tenuat ion factor to permit correct ion for at tenu- at ion during tomographic reconstruction. Consequently we have chosen to perform reconstruct ion without a t tenuat ion

Fig. 4. Arthrogram a and tomographic slices (sagittal h, transaxial e and coronal d) obtained from the patient with highest 9Oy uptake in a Baker's cyst (patient 2). The approximate planes of the transaxial (T) and coronal (C) slices are shown in a

502

Table 1. Estimated 9oy content of Baker's cysts (MBq) and reten- tion (% of administered activity)a at different times

Patient 1-2 h 24 h 46 h

MBq % MBq % MBq %

1 27.0 14.6 19.3 10.4 20.8 11.2 (17.6) (12.6) (13.5)

2 68.9 37.2 74.6 40.3 65.8 35.5 (34.1) (36.6) (32.3)

3 0 0 0 0 0 0

4 3.2 1.7 l lA 6.0 8.9 4.8 (1.4) (4.9) (4.1)

5 1.3 0.7 5.0 2.7 6.3 3.4 (0.6) (2.3) (2.9)

6 27.9 15.1 36.3 19.6 30.9 16.7 (12.7) (16.5) (14.7)

7 0 0 0 0 0 0

8 14.4 7.8 13.6 7.3 12.6 6.8 (6.2) (5.9) (5.5)

9 22.0 11.9 not done 18.8 10.2 (11.6) (9.6)

tO 1.9 1.0 6.8 3.7 4.1 2.2 (1 A) (3.9) (2.4)

a First retention value is calculated from the activity content of the cyst assuming an administered activity of 185 MBq 9oy. Value in parentheses is the ratio of decay corrected counts in the cyst to the initial counts in the total knee

correction and resort to the use of realistic phantoms for calibration purposes. In these studies, a 10% threshold value has been used for comparing count rates in order to maintain a high counting efficiency within the region of interest (Fig. 2 a) while limiting the effects of noise due to reconstruction artefacts and background.

From the data obtained with calibration phantoms (Fig. 2b) there was apparently little influence due to the presence o f bones (approximately 3%-10%), the larger ef- fect occurring with smaller source volumes. However, direct comparison between the two sets of data, with and without bones, is complicated by the fact that source positions were not the same in each case, being of necessity offset from the central axis when bones were included. The effect o f attenuation of 90y Bremsstrahlung is illustrated by the in- crease in counting efficiency with increase in source volume and with decrease in phantom diameter.

For accurate quantification o f cyst activity using cali- bration data from phantoms it was necessary to estimate cyst volume. This was done by computer calculation of the number of voxels (0.25 cm 3) enclosed within a counting threshold selected in relation to the maximum pixel count within a region of interest which isolated the cyst from regions of higher activity in its proximity. In this we experi- enced the same difficulty as reported, for example, by Tauxe et al. (1982) in the choice o f the appropriate threshold to determine the volume of a given cyst. Fig. 2c illustrates the need for accurate thresholding and gives some indica- tion of the errors that could arise. Ott et al. (1987), using positron emission tomography with 124i, observed a change in slope o f the volume threshold curve to predict the appro-

Fig. 5. Variation in 9oy distribution in the knee of patient 6 during the first 24 h

priate threshold value for thyroid volume estimation, but we were unable to demonstrate a similar effect with our 9oy phantoms. However, as Fig. 2d indicates, the appro- priate threshold is close to 50% for volumes greater than about 30 cm 3. A spherical cyst of approximate radius 2 cm has this volume and we have used a 50% threshold to esti- mate cyst volume, or more correctly the volume of a cyst occupied by 90y, in our patients.

503

Serial measurements on knees showed variable retention of 9oy ranging from 62% to 98% at 2 days poatinjection. As observed by others (Williams 1973), the deficit in knee activity was in most cases only partially accounted for by uptake in liver and lymph nodes. It is possible that, in some cases, the changes in distribution of 9oy in the knee might explain part of the discrepancy. The calibration data of Fig. 2b show that counting efficiency fails as the source volume is reduced, and situations such as that shown in Fig. 5, in which the 9Oy distribution contracts with time, could lead to erroneously low measurements of retention. Redistribution is possibly due to slow settlement of 9oy colloidal particles and hence related to particle size (Ingrand 1973). In this particular patient, for whom the largest 2 day fall in knee retention (38%) was observed, it is esti- mated that the effect of redistribution artefacts could amount to 10%-15% of the administered activity. Liver uptake in this case was only 3% and it is presumed that the residual activity unaccounted for is generally distributed throughout the body.

The clinical implications of the results of these studies (Table 1) have been discussed elsewhere (Smith et al. 1988). In summary, they show that uptake of 9oy in cysts was:

(1) extremely variable ( 0 % 4 0 % in this small group of 10 patients);

(2) not obviously related to the size of the cysts as deter- mined from arthrograms, and

(3) not a reliable indicator of attenuated therapeutic response as observed at 6 and 12 months follow up.

These results suggest that, in our patients, variations in dose determining factors such as spatial relationships between 9oy particles and target cells, and in the radio- sensitivity of the synovial membrane, were of greater conse- quence in terms of therapeutic response than the effects of uptake of some 9oy in Baker's cysts. Thus, on the evi- dence of these results we were unable to propose a scheme whereby the magnitude of the administered activity might ba varied, according to the size of Baker's cysts on arthro- grams, to compensate for the anticipated removal of 9Oy from the synovial space by uptake into cysts.

In conclusion, it has been shown that, for a relatively thin region of the body, such as the knees, SPECT studies can be performed satisfactorily using pure fl emitting radio- nuclides, such as 9oy, whose beta particles are sufficiently energetic to allow efficient detection of their Bremsstrah- lung radiation by gamma camera. In this particular applica- tion, the ability to examine mutually perpendicular slices (0.63 cm thick) through the knee region (Fig. 4) provided a distinct advantage over planar imaging in aiding discrimi- nation between activity in cysts and that present in adjacent sites such as the synovial capsule and supra patellar pouch. For applications involving deep seated activity loci within thicker parts of the body the effect of attenuation on the continuous Bremsstrahlung spectrum would be more acute, leading to greater problems due to low sensitivity, poor resolution and difficult calibration.

References

Baker WM (1877) The formation of synovial cysts in the leg in connection with disease of the knee joint. St. Bart's Hosp Rep t 3 : 245-261

Balachandran S, McGuire L, Flanigan S, Shah H, Boyd CM (1985) Bremsstrahlung imaging after 3ap treatment for residual supra- sellar cyst. Int J Nucl Med Biol 12:215-221

Boye E, Lindegaard MW, Paus E, Skretting A, Davy M, Jakobsen E (1984) Whole-body distribution of radioactivity after intra- peritoneal administration of 32p colloids. Br J Radiol 57:395M02

Crawley JCW, Smith T 0986) Some problems of quantitative up- take measurements with single photon emission tomography. In : Clark RP, Goff MR (eds) Recent developments in medical and physiological imaging. Taylor and Francis, London, p 147

Croll MN, Brady LW (1979) Intracavitary uses of colloids. Semin Nucl Med 9 : 108-113

Gumpel JM, Williams ED, Glass HI (1973) Use of yttrium-90 in persistent synovitis of the knee. Ann Rheum Dis 32 : 223-227

Ingrand J (1973) Characteristics of radio-isotopes for intra-articu- lar therapy. Ann Rheum Dis [Suppl] 32: 3

Kaplan WD, Zimmerman RE, Bloomer WD, Knapp RC, Adel- stein SJ (1981) Therapeutic intraperitoneal 32p: a clinical as- sessment of the dynamics of distribution. Radiology 138:683-688

Kobayashi T, Kageyama N, Ohara K (1981) Internal irradiation for cystic craniopharyngioma. J Neurosurg 55:896-903

McGuire EL, Balachandran S, Boyd CM (1986) Radiation dosi- metry considerations in the treatment of cystic suprasellar neo- plasms. Br J Radiol 59 : 779-785

Ott RJ, Batty V, Webb S, Flower MA, Leach MO, Clack R, Mars- den PK, McCready VR, Bateman JE, Sharma H, Smith A (1987) Measurement of radiation dose to the thyroid using posi- tron emission tomography. Br J Radiol 60:245 251

Pezner RD, Stevens KR, Tong D, Allen CV (1978) Limited epithe- lial carcinoma of the ovary treated with curative intent by the intraperitoneal installation of radiocolloids. Cancer 42:2563- 2571

Smith T, Shawe DJ, Crawley JCW, Gumpel JM (1988) Use of single photon emission computed tomography (SPECT) to study the distribution of 9oy in patients with Baker's cysts and persistent synovitis of the knee. Ann Rheum Dis 47:553 558

Spooren PFMJ, Basker JJ, Arens RPJH (1985) Synovectomy of the knee with 9oy. Eur J Nucl Med 10:441-445

Sullivan DC, Harris CC, Currie JL, Wilkinson RH, Creasman WT (1983) Observations on the intraperitoneal distribution of chromic phosphate (32p) suspension for intraperitoneal thera- py. Radiology 146:539 541

Tauxe WN, Soussaline F, Todd-Pokropek A, Cao A, Collard P, Richard S, Raynaud C, Itti R (1982) Determination of organ volume by single-photon emission tomography. J Nucl Med 23 : 984~987

Taylor AR (1969) Arthrography of the knee in rheumatoid arthri- tis. Br J Radiol 42:493-497

Williams ED (1973) methods of measuring retention of radioiso- topes in the knee and extra-articular spread in vivo. Ann Rheum Dis [Suppl] 32:13

Received January 16, 1988