2019 Research Grant

 

 

Geoff Currie and Stephanie Sanchez, Charles Sturt University

 

RESEARCH  

The aims of this project were to provide an Australian context to extravasation rates in PET using the quality assurance device (LARA produced by Lucerno Dynamics and distributed in Australia by Bayer).

As previously outlined, PET and SUV depend on reliable pharmacokinetic modelling, part of which is predictable dose delivery. Partial extravasation of the intravenous (IV) dose administration undermines predictability of dose delivery and potentially the accuracy of SUV calculation. The device (LARA) is a simple, non-invasive way to not only determine partial dose extravasation but to potentially characterise the pharmacokinetic behaviour of the extravasated volume. Given the value of SUV in patient management, determining treatment options and monitoring progress or response to treatment, this form of quality assurance represents an essential and powerful tool in the clinical PET department. The extravasated portion of doses may go undetected if the injection site is outside the field of view and standard imaging protocols with the arms extended above the head ensure this is the case. Data from the USA using this device suggests as many as 33% of doses are partially extravasated and go undetected.

The work resulted in 4 peer reviewed journal publications:

Sanchez, S & Currie, G 2020, Topical sensor for the assessment of injection quality for 18F-FDG, 68Ga-PSMA and 68Ga-DOTATATE positron emission tomography, Journal of Medical Imaging and Radiation Sciences, vol. 51; no. 2: pp. 247-255. https://doi.org/10.1016/j.jmir.2020.01.003

This first publication provided an evaluation of the extravasation rates at the first site of data collection for 18F FDG, 68Ga PSMA and 68Ga DOTATATE. This extended previous understanding because the LARA device had only previously been validated with 18F FDG. In total, 296 patients were evaluated using the proprietary metrics. The extravasation rate was only 1.1% which was substantially lower than previously reported rates. The analysis also identified 9.1% of patients with abnormal time activity curves or metrics (score) in the absence of extravasations. That is, slow venous clearance of the injection. The study concluded that topical monitoring and characterisation of PET dose administration is possible and practical with the LARA device which could provide the insights into variables that could eliminate extravasation as a cause of image quality or SUV accuracy issues.

Sanchez, S & Currie, G 2020, Topical Sensor for the Assessment of Positron Emission Tomography dose extravasation; Metric Performance with an Autoinjector, Journal of Nuclear Medicine Technology, in press, doi:10.2967/jnmt.120.245043

The data was further evaluated form a metric performance perspective. The “score” used to guide decision making is based on weighted product of a number of simple metrics. This paper outlines the calculation of these metrics and the potential limitations. Of note was that the majority of the metrics depended on the actual dose administered and presumed via original validation a 18F FDG dose of approximately 600MBq. Consequently, the metrics contained errors when used for PET scans with doses more typically 220 MBq. A number of specific metrics were demonstrated to be independently more predictive of injection quality including TC50 and ndAvgN. We introduced our own metric (CEnd ratio) which normalised the total count dependence of data at the end of the monitoring period and this was also shown to be robust. The paper concluded that partial extravasation of PET doses can be readily detected and differentiated using time activity curve metrics. The paper also recommended validation of key metrics in a larger and more diverse cohort.

Sanchez, S & Currie, G 2020, Topical device detection of 18F FDG dose leakage, Journal of Nuclear Medicine Technology, vol. 48; no. 3: pp. 283-284, doi:10.2967/jnmt.119.240283

This paper is a case study from the primary data collection that provided a unique insight into data interpretation. Indeed, it highlighted the critical importance of manual evaluation of time activity curves rather than reliance on proprietary metrics alone. In this case, the LARA data indicated a severe extravasation of the injection and the PET scan image quality supported that position. Closer investigation, however, revealed that the dose had not been extravasated but rather that the connector was leaky and the dose leaked into the pillow on which the injection arm was being supported. The teaching case concluded that auto-injector set-up with a large bore cannula and high volume flush minimise partially extravasated doses compared to hand injection and reduce staff radiation dose but leave the patient and injection unsupervised. The LARA device provided a useful tool for more timely critical evaluation and problem solving; extending advantage to the patient and practice.

Currie, G & Sanchez, S 2020, Topical Sensor metrics for 18F-FDG positron emission tomography dose extravasation, Radiography, in press, https://doi.org/10.1016/j.radi.2020.07.013

Following the results from the primary site, a second site was recruited to collect patient data. Unfortunately COVID19 interrupted data collection at the 100 patient mark. A multi-site, multi-national pooling of 18F FDG PET/CT data was undertaken with 863 patients from 6 sites in the USA and 2 sites in Australia. A number of metrics from LARA were analysed using conventional statistical analysis and using an artificial neural network. The results indicated that extravasation was independently predicted by the time taken for the injection sensor counts to reach double the counts of the reference sensor (tc50), the normalised difference between injection and reference sensors counts at 4 minutes post injection (ndAvgN), or the ratio of injection sensor counts to reference sensor counts at the end of data collection (CEnd ratio). The algorithm developed using the artificial neural network produced 100% sensitivity and 100% specificity against grounded truth for detecting extravasation by weighting and scaling these 3 key metrics; CEnd ratio, ndAvgN and tc50. The data also revealed a number of significant differences between the Australian data and the USA data that could explain the lower extravasation data reported in some departments (table 1).

Table 1: A summary of the combined demographic data and sub-divided by country. The bold data reflects statistically significant difference between the USA and Australian cohorts for key variables.

	USA only (n=569)	Australia only (n=294)	P
Proportion of studies (%)	65.9	34.1
Extravasation rate (%)	8.8	1.0	<0.001
Severe extravasation rate (%)	1.9	0.3	<0.001
Venous stasis (%)	6.3	0.7	<0.001
Mean dose (MBq)	568.9 (95% CI 562.9-574.9)	219.8 (95% CI 211.5-228.1)	<0.001
Over 70 years (%)	41.8	48.3	0.151
Flush (mL)	9.5 (95% CI 8.9-10.2)	62.3 (95% CI 60.5-64.1)	<0.001
Autoinjector (%)	2.8	99.0	<0.001
20 gauge (or larger) cannula (%)	2.0	77.9	<0.001
Antecubital injection (%)	56.9	83.3	<0.001
Left side injection (%)	30.6	70.4	<0.001
Butterfly injection (%)	46.2	0	<0.001
Mean tc50	333.9 (95% CI 283.6-384.3)	116.2 (95% CI 46.0-186.4)	<0.001
Mean ndAvgN	18.0 (95% CI 9.4-26.5)	0.6 (95% CI -11.2 to 12.4)	0.020
Mean CEnd ratio	2.1 (95% CI 1.2-2.9)	1.0 (95%CI -0.2 to 2.2)	0.169

 

Table 2: Summary of metrics against outcome of injection (combined data only).

	Normal	Abnormal but insignificant	Vascular Retention	Total Extravasated	Severe Extravasated	P
Mean dose (MBq)	423.9	487.3	561.2	568.6	484.4	<0.001
Over 70 years (%)	43.2	43.4	44.4	65.7	50.0	0.182
Injection in hand (%)	19.0	35.1	7.9	43.9	16.7	<0.001
Left side injection (%)	50.2	30.4	29.0	29.3	33.3	<0.001
Butterfly injection (%)	27.7	29.8	52.6	48.8	50.0	<0.001
Mean flush volume (mL)	31.0	23.2	12.0	13.1	12.7	<0.001
Autoinjector used (%)	42.6	26.8	5.3	4.9	8.3	<0.001
20 gauge or larger needle (%)	33.1	21.0	5.2	4.9	18.2	<0.001
tc50	52.8	377.8	412.6	2082.8	2302.6	<0.001
ndAvgN	-0.5	9.2	11.3	47.7	564.3	<0.001
CEnd ratio	0.97	1.23	1.03	2.74	43.14	<0.001

 

Discussion

Extravasation rates and the severity of extravasation can be minimised by:

• More careful management of injections in older patients.
• Where possible, injecting in the antecubital fossa.
• Use of a cannula and avoid butterfly administrations.
• Use of a 20 gauge (or larger) bore for the injections.
• Use of an auto-injector instead of manual injection (avoiding rapid bolus and ensuring higher flush volume).

The recognition of extravasated doses as misadministration with associated investigation and reporting requirements of ARPANSA in Australia may also be a factor in reducing rates through training and education. In contrast, the NRC in the USA, with respect to radiopharmaceutical administration, have determined that “extravasation frequently occurs ……… it is virtually impossible to avoid …….. does not consider extravasation to be misadministration…” and thus exempts reporting and accountability requirements in cases of extravasation misadministration.

Extravasation was independently predicted by a tc50 and ndAvgN. Clearly these metrics lack the ability to discriminate between normal, vascular retention and extravasated injections for some patients. There is potential to improve the robustness of metrics by including the CEnd ratio and limiting aUCR10 to the window 4-10 minutes post injections (aUCR_4-10) instead of 1-10 minutes. Mathematical calculation of the 3-minute peak-to-tail ratio (PT3 ratio) and the extravasation fraction is a potentially important metric in determining impact and recourse associated with SUV calculation. This can be automated to allow the user to select the start points of both region A and region B, windowed for 3 minutes (180 seconds), subtraction from each value the corresponding reference count, summation of all points in each 3 minute window and dividing the value for A by B to produce a ratio. The inverse of that expressed as a percentage could be expressed as the extravasation fraction. These are outlined in detail in the 4^th publication (Currie & Sanchez 2020).

Conclusion

Partial extravasation of PET doses can be readily detected using topical sensors or imaging the injection site during scanning. Partial extravasation of the PET dose is inadequate and threatens both the integrity of the study quantitation and the procedural reputation. Extravasation should require reporting and should demand training / education to ensure reliability of SUV, optimal image quality, improved patient outcomes and limitation of unnecessary tissue doses to patients. The techniques and metrics developed in this research provide a cheap and simple mechanism to detect extravasation and characterise extravasation severity; potentially providing a mechanism to adjust dose delivered and SUV accuracy.