Assessment of early treatment response after CT-guided radiofrequency ablation of unresectable lung tumours by diffusion-weighted MRI: a pilot study

Abstract

The aim of this study was to evaluate prospectively the early treatment response after CT-guided radiofrequency ablation (RFA) of unresectable lung tumours by MRI including diffusion-weighted imaging (DWI). The study protocol was approved by the ethics committee of our hospital and signed consent was obtained from each patient. We studied 17 patients with 20 lung lesions (13 men and 4 women; mean age, 69±9.8 years; mean tumour size, 20.8±9.0 mm) who underwent RFA using a LeVeen electrode between November 2006 and January 2008. MRI was performed on a 1.5T unit before and 3 days after ablation. We compared changes in the apparent diffusion coefficient (ADC) on DWI and response evaluation based on subsequent follow-up CT. 14 of the 20 treatment sessions showed no local progression on follow-up CT, whereas 6 treatment sessions showed local progression (range, 3–17 months; mean, 6 months). For the no-progression group, the ADC pre- and post-RFA were 1.15±0.31 × 10⁻³ mm² s⁻¹ and 1.49±0.24 × 10⁻³ mm² s⁻¹, respectively, while the respective ADC values for those that showed local progression were 1.05±0.27 × 10⁻³ mm² s⁻¹ and 1.24±0.20 × 10⁻³ mm² s⁻¹. The ADC of the ablated lesion was significantly higher than before the procedure (p<0.05). There was a significant difference in the ADC post-RFA between no-progression and local progression groups (p<0.05). Our prospective pilot study showed that the ADC without local progression was significantly higher than with local progression after RFA, suggesting that the ADC can predict the response to RFA for lung tumours.

After the first report in 2000 [1], lung radiofrequency ablation (RFA) is now considered effective in the treatment of lung cancer, which is traditionally considered unresectable owing to compromised pulmonary function or advanced age. In general, complications associated with lung RFA are minimal, and favourable local control has been reported in a number of studies of tumours with a diameter of 30 mm or less [1–5]. However, only a limited number of studies have been published regarding the treatment outcome after lung RFA [6–10]. In this process, a layer of normal lung tissue surrounding the tumour is also ablated as a safety margin. Inevitably, the ablated lesion depicted on a CT scan immediately after the procedure is larger than the original tumour mass. However, this region of increased density shrinks with time, but follow-up CT may still show the ablated lesion being as big as, or larger than, the tumour size before the procedure [6, 7]. Thus, radiologists sometimes encounter difficulty in distinguishing scarred tissue from a tumour residue/local progression when the size of the lesion remains the same. Accurate assessment of RFA outcome would have important consequences, as recurrent tumours can be treated again if detected at an early stage. Different modalities of early-stage follow-up examination, such as contrast-enhanced CT [8] and fluorodeoxyglucose positron emission tomography (FDG–PET), have been of great interest and their usefulness has been reported by several groups [9, 10]. Another approach — MR diffusion-weighted imaging (DWI) — which is based on the measurement of motion of water molecules, has also been reported as a non-invasive evaluation modality [11–19]. In this method, the apparent diffusion coefficient (ADC) represents the water content and distribution, the cellular density and the cell membrane integrity, suggesting the potential usefulness of an ADC map for estimating tumour viability. Indeed, DWI has been successfully used to assess the efficacy of radiotherapy [11, 12], chemotherapy [13–15] and transcatheter arterial embolisation [16, 17]. To our knowledge, only two studies have reported the use of DWI to evaluate the therapeutic outcome of RFA [18, 19]. A previous study reported that the ADC value of an ablated rabbit tumour model (VX2 tumour) was significantly higher than that of untreated tumours, and that FDG uptake on micro-PET for small animals with ablated tumours was significantly lower than for untreated tumours. These results indicate that DWI at 2 days and FDG–PET at 3 days after RFA are both potentially feasible modalities for monitoring the early effects of the procedure [19]. In this study, we calculated the ADC in tumour lesions before and after clinical lung RFA and examined the usefulness of DWI in the early detection of tumour response to RFA.

Methods and materials

Patient background

The study protocol was approved by the ethics committee of our hospital and signed consent was obtained from each patient. Lung RFA was performed in 24 consecutive treatment sessions on 21 patients (17 males and 4 females; mean age, 71±9.8 years) with unresectable lung cancer between November 2006 and March 2008. Of these, four treatment sessions in four patients, with follow-up CT of less than 3 months, were excluded. This study progressed with 17 patients (13 males and 4 females; mean age, 69±9.8 years; 20 sessions) with lung cancer. In these patients, RFA was performed for primary lung cancers (n _ 10 treatment sessions), as well as metastatic tumours from colorectal (n _ 6) and cervical cancer (n _ 1). The mean tumour size was 20.8±9.0 mm (range, 10–45 mm).

Lung RFA technique

As reported previously [20], the indication for lung RFA is inoperable lung tumour(s) owing to previous surgical history, compromised pulmonary reserve, the presence of multiple lesions or other reasons after discussion with thoracic surgeons. Patients with a primary lung cancer were diagnosed based on pathological examination obtained by the easiest method (transbronchial needle biopsy or fine needle aspiration). Pulmonary metastases were diagnosed based on clinical course, imaging studies and haematological data. The detailed procedure of RFA of lung tumours and the inclusion/exclusion criteria have been described previously [20]. Briefly, the electrode needle was inserted after local anaesthetisation of the subcutis through the peripleural tissue, and positioned toward the targeted area under CT guidance. The needle was deployed to start ablation. For tumours ≤20 mm in size and those >20 mm, a needle with a 20 mm and 30 mm expandable tip, respectively, was used. We used an RF 2000 generator and LeVeen electrodes (Boston Scientific Corporation, Natick, MA). Initially, the RF power was set at 20 W for tumours <20 mm and 30 W for tumours ≥30 mm, and then increased by 5 W at 2 min intervals. The maximum RF power was 80 W. Ablation was completed at “roll-off”, at which point impedance reaches maximum and RF is automatically shut off. To ablate the entire tumour mass, power was applied several times (average, 2.7±1.1 times; range, 1–5 times per lesion) for a single mass (so-called “overlapping ablation”).

MRI sequence

MRI was performed 4 days before and 3 days after RFA at 1.5T using Magnetom Avanto (Siemens Medical Solutions, Erlangen, Germany) with (i) breath-holding for spin-echo T₁ weighted images (8 mm thick, interslice gap 2 mm, field of view (FOV) _ 38 cm, repetition time/echo time (TR/TE) _ 40/4.7 ms, matrix size 256 × 208), fast-spin-echo T₂ weighted images (8 mm thick, interslice gap 2 mm, TR/TE _ 3800/82 ms, matrix size 256 × 208) and T₂ weighted half-Fourier acquisition single-shot turbo spin-echo (HASTE-T₂; 8 mm thick, interslice gap 2 mm, TR/TE _ 1300/92 ms, matrix size 256 × 208), and (ii) quiet breathing for short-time inversion recovery (STIR)-DWI (6 mm thick, gapless, TR/TE _ 5700/82 ms, b-values 0 and 1000, matrix size 128 × 128) and chemical shift selective saturation (CHESS)-DWI (TR/TE _ 81/5700 ms, b-values 0, 500 and 1000, matrix size 128 × 128).

CT

For follow-up CT, a CT scanner with four rows of detectors (multidetector row CT) was used (Asteion multi, Toshiba, Tokyo, Japan). Scanning parameters were 120 kVp, 220 mA, 2 mm section collimation and 7 mm image reconstruction. CT was performed within 1 week after RFA, at 3, 6 and 12 months after RFA, and then every 6 months to examine local progression. The therapeutic outcome was determined using CT scans taken 3 months after RFA or later. In seven cases, follow-up CT could not be performed 3 months after RFA, and thus only the remaining 14 patients (11 males and 3 females; 8 with primary lung tumours, 5 with metastatic tumours from colorectal cancer, 1 from cervical cancer, and 1 from oesophageal cancer; 17 sessions; median age, 70±7.9 years) were determined for therapeutic outcome.

Imaging analysis

Two radiologists (T.O. and M.T.) examined the MR and non-contrast CT images and calculated ADC from ADC maps generated from the CHESS-DWI data. The transverse T₁ weighted, T₂ weighted and STIR-DWI images were compared visually with muscle and graded as hypo-, iso- or hyperintense. The MR data were reconstructed in a Siemens workstation (SINGO), and ADC maps were generated from the DWI data. Using T₁ and T₂ weighted images as references, the region of interest (ROI) was set at the centre of the largest tumour section to calculate the ADC. When assessing the outcome by CT, the most recent CT images were compared with those immediately after RFA, not before RFA; they were considered to have responded when the ablated legion shrank over time and they were considered to show local progression when the lesion was larger or showed changes in shape such as protrusion or an irregular, scattered, nodular or eccentric focus from the margin [6]. CT images before and after the procedure were reviewed, and decisions were reached by consensus.

Statistical analysis

Statistical analysis was performed by the paired t-test using commercially available software (Prism 4 for Macintosh; GraphPad Software, Inc., San Diego, CA). We compared (i) ADC values before and after RFA, (ii) pre- and post-RFA ADC values between patients with local progression and those without, (iii) pre-RFA ADC values between the two groups, (iv) post-RFA ADC values between the two groups, (v) pre-RFA tumour size between the two groups, (vi) correlation of pre- and post-RFA ADC values with tumour size, and (vii) correlation of pre- and post-RFA ADC values with primary pulmonary lesions and secondary lesions. Data are expressed as mean±SD (standard deviation). A p-value of <0.05 was considered statistically significant.

Results

In all of the 20 treatment sessions, the signal intensity of the lesion was similar before and after RFA on T₁ weighted sequences. On T₂ weighted images, tumours appeared as areas of slightly higher signal intensity compared with the muscle tissue before treatment; however, after RFA, the signal intensity became lower at the centre and higher in the peripheral zone. The lesion size appeared larger on both T₁ and T₂ weighted sequences after treatment. Post-RFA DWI showed reduced signal intensity compared with pre-RFA images in all cases (Figure 1). The mean ADC value of tumours was 1.15±0.27 × 10⁻³ mm² s⁻¹ before RFA, and increased significantly to 1.49±0.24 × 10⁻³ mm² s⁻¹ after RFA (p<0.01; Figure 2).

Figure 1.

A 79-year-old man with a pulmonary metastatic lesion in the left upper lobe (a–c) before and (d–f) after radiofrequency ablation (RFA). (a) Transverse T₁ weighted MR image (repetition time/echo time (TR/TE), 40/4.7 ms) shows an isointense mass (arrow) relative to the surrounding muscles. (b) T₂ weighted image (TR/TE, 3800/82 ms) shows a hyperintense mass (arrow) relative to muscle. (c) Diffusion-weighted image (TR/TE, 5700/82 ms) taken before RFA shows a slightly hyperintense mass (arrow). The apparent diffusion coefficient (ADC) value before RFA was 1.44 × 10⁻³ mm² s⁻¹. (d) On a T₁ weighted MR image obtained 3 days after RFA, the ablated lesion (arrow) is isointense, similar to before the procedure. (e) T₂ weighted MR image shows an inner hypointensity area with outer hyperintensity area (arrow). (f) Diffusion-weighted image (TR/TE, 5700/82 ms) after RFA shows decreased signal intensity (arrow), compared with images obtained before the procedure. The ADC value after RFA was 1.76 × 10⁻³ mm² s⁻¹.

Figure 2.

Comparison of pre- and post-radiofrequency ablation (RFA) apparent diffusion coefficient (ADC) values in tumours with and without subsequent local progression. There was a significant difference between pre- and post-treatment ADC values in six treatment sessions that subsequently showed local progression (p<0.05). Furthermore, pre- and post-RFA ADC values were significantly different in 14 treatment sessions that did not show local progression. (p<0.01). However, there was no significant difference between pre-RFA ADC values with or without local progression. The post-treatment ADC of lesions with local progression was significantly different from that of lesions without local progression (p<0.01). Data are given as mean±SD (standard deviation). NS, not significant.

During the follow-up period, local progression was detected in 6 treatment sessions between 3 months and 17 months after RFA. For the treatment sessions that showed local progression, the mean ADC was 1.05±0.27 × 10⁻³ mm² s⁻¹ before RFA, and increased significantly to 1.24±0.20 × 10⁻³ mm² s⁻¹ after RFA (p<0.05). In treatment sessions that did not show local progression, the mean ADC was 1.15±0.27 × 10⁻³ mm² s⁻¹ before RFA, and increased significantly to 1.49±0.24 × 10⁻³ mm² s⁻¹ after RFA (p<0.01). There was also a significant difference in the post-RFA ADC values between the treatment sessions with local progression and those without (p<0.05), but not in the pre-RFA ADC values (p _ 0.46). The tumour size of the treatment sessions with local progression (30.8 ± 7.9 mm) was significantly larger than that of the treatment sessions without local progression (16.5±5.1 mm; p<0.01; Figure 2). The correlation between pre- and post-ADC values of lesions and the tumour size before the procedure was not significant (p _ 0.25 and p _ 0.08, respectively). There were no significant differences in primary pulmonary and secondary lesions pre- and post-RFA (1.09±0.06 × 10⁻³ mm² s⁻¹ vs 1.17±0.12 × 10⁻³ mm² s⁻¹ pre-RFA, p _ 0.49; and 1.39±0.07 × 10⁻³ mm² s⁻¹ vs 1.45±0.10 × 10⁻³ mm² s⁻¹ post-RFA, p _ 0.63)

Discussion

RFA for lung tumour is minimally invasive and the associated complications are generally minor; however, follow-up protocols and criteria for monitoring treatment response after the procedure remain controversial [1–10]. We evaluated the use of MRI to predict the treatment response to RFA for lung tumours. DWI at 3 days after lung RFA showed reduced signal intensity and significantly increased ADC values at the ablated lesions compared with pre-operative tumour tissues. There was also a significant difference in the mean ADC of lesions with and without local progression. These results suggest that pre- and post-RFA DWI signals can be used to predict the therapeutic outcome before the change in tumour size becomes detectable on CT images. Furthermore, even a little increase in the ADC value after RFA should suggest that local progression is likely to occur in future follow-up CT scans.

The diffusion characteristics of protons reflect biological parameters such as cell density and nuclear volume fraction in the tumour tissue, and a decrease in cell density or nuclear volume fraction results in an overall increase in the ADC. Only a few studies have used DWI for assessment of treatment outcome and local progression, but DWI has been used after chemoradiotherapy for brain tumours [11, 12], chemotherapy for breast and bone cancers [13–15] and transarterial chemoembolisation for hepatic tumours [16, 17]. In these studies, ADC values have been shown to increase after treatment, and to have a good predictive value for therapy outcome. The mechanism of increased ADC after anticancer treatment is unclear, but is probably associated with necrosis, which results in cellular shrinkage and decreased intracellular water content [13]. In the present study, we observed an increase in tumour ADC after treatment, a finding consistent with other studies [11–17]. Only a few studies have reported the use of DWI after RFA [18, 19, 21]. One such study examined the usefulness of MRI and FDG–PET for assessment of the effect of RFA in rabbit models with VX2 tumour implanted in the muscle of the back. The authors reported an increase in tumour ADC at 2 days post-RFA, and the pathological findings included ongoing necrosis of tumour cells, decreased cell density and increased extracellular space [19]. Another group used DWI to evaluate the outcome within 1 week of RFA in patients with lung cancer [18]. In that study, no signal was detected after successful treatment, but the signal was still present in incompletely ablated lesions. Furthermore, the authors performed RFA in tumour-bearing nude mice, and also reported no signal on DWI images at 2 h after treatment [18]. In the present study, we also observed a reduction of signal intensity 3 days after lung RFA, consistent with their findings. In the above study [18], however, the signal intensity was studied only visually, and no quantitative analysis, such as ADC, was performed. Conversely, the signal intensity did not change before and after RFA on T₁ weighted sequences, but showed a central hypointense zone with a peripheral zone of higher signal intensity on T₂ weighted images. In another study that examined post-RFA MRI in normal lung tissues, the inner zone showed hypointensity on T₂ weighted images and isointensity on T₁ weighted images, corresponding to coagulative necrosis. The peripheral zone demonstrated hyperintensity on T₂ weighted images, representing neutrophilic infiltration, intra-alveolar fluid collections and pulmonary congestion, which was seen on pathological examination in a porcine model. The T₁ and T₂ signal pattern observed in the present study was consistent with their results [21]. On T₁ weighted images, however, they observed equivalent isointensity in the ablated area, which hindered accurate assessment. They also concluded that the low signal area underestimates the size of the thermal lesion [21]. Considered together, the above studies and the present results indicate that T₁ and T₂ weighted imaging alone is not sufficient for assessment of RFA outcome.

The timing of post-operative DWI varies from 1 day to 8 weeks depending on the study [11–19, 22–25]. Several studies concluded that DWI is informative when conducted several days after treatment [11, 13, 18, 19, 22, 23]. In our study, DWI was performed 3 days after RFA to examine its predictive value for therapy outcome. After RFA, a significant increase in ADC was observed, and post-treatment ADC was significantly lower with future local progression. In addition, the mean tumour diameter was significantly larger in lesions that subsequently showed local progression; this result is consistent with previous lung RFA studies that have concluded that a tumour diameter larger than 30 mm is significantly more frequently associated with local progression [3–5].

The limitations of this study are the relatively small number of subjects, relatively short follow-up period, thickness of the MRI scan and inclusion of various types of lung tumours. The ROI was delineated at the centre of the ablated tumour, and therefore local progression outside the ROI was not counted. Youn et al [25] argued that, owing to the limited resolution of MR images, tumours are not detectable on DWI when they are too small. An 8 mm slice thickness with a 2 mm interslice gap might miss a small nodular local progression or make interpretation difficult owing to partial volume. Because “local progression” was defined on follow-up CT by the size of tumour compared with the reference images, MRI resolution should not affect the diagnosis of local progression. Conversely, patients without local progression may show local progression in future CT scans, because the follow-up period was too short in this study.

In conclusion, in lung RFA, the tumour ADC value calculated from DWI performed 3 days post-treatment was useful for predicting therapy outcome, and it may allow early assessment of tumour response before changes in tumour size become detectable on CT.