The effect of selective internal radiation therapy with yttrium-90 resin microspheres on lung carbon monoxide diffusion capacity

Selective internal radiation therapy (SIRT), in which the branches of the hepatic artery are embolized using biocompatible Yttrium-90 (⁹⁰Y) labeled microspheres, have recently emerged as an important therapeutic tool for patients with hepatic malignancy. However, after intraarterial injection of ⁹⁰Y-labeled microspheres into the liver, a substantial portion of the radioactive material is shunted into the lung via intrametastatic/intratumoral arteriovenous shunts [1]. When proportion of the shunting of radionuclide microspheres exceeds 15%, the risk of radiation induced pneumonitis (RP) is significantly elevated [2]. It is recommended that SIRT is relatively contraindicated in patients with lung shunting such that the single treatment lung dose will exceed 30 Gy or that the cumulative lung dose will exceed 50 Gy [3, 4].

Diffusion capacity of the lungs for carbon monoxide (DLCO), a clinically important measure of the lung function, gauges the ability of the lungs to transfer the gas in inhaled air to the red blood cells in pulmonary capillaries [5]. General areas of use include the identification of the cause of dyspnea or hypoxemia, monitor the progression of interstitial lung disease, and to detect the presence of pulmonary hypertension in patients under risk [6]. In addition, DLCO may also show the presence of limited gas exchange reserve caused by the potential toxicity of chemoradiotherapy, and previous studies have suggested that the most consistent changes in pulmonary function tests after external beam radiotherapy (RT) are recorded with DLCO [7–9].

Radiation pneumonitis is a rare but serious complication of SIRT that may develop 1 to 6 months after treatment. Further radiation exposure in lungs is not recommended within the first 6 months after SIRT, and in cases where the second SIRT is performed within 6 to 12 months of initial treatment, the recommended dose is < 50% of the maximum tolerated dose that was administered during the initial procedure [10, 11].

Also, DLCO may be a sensitive marker of lung injury, despite the reduction in DLCO is usually subclinical [9, 12–15]. Published data on the degree of radiation-related changes in DLCO after SIRT is limited, and the importance of subclinical lung injury in these patients remains unknown. Nevertheless, DLCO has been recommended as part of the general work-up of such cases [11].

This study aimed to examine the changes in DLCO during SIRT with resin microspheres in newly treated and retreated patients.

The mean DLCO value did not significantly change after the first (82.8 mmol/min/kPa ± 19.4 vs. 83.1 ± 20.9, p = 0.921) and the second treatments (87.4 ± 19.7 vs. 88.6 ± 23.2, p = 0.256). In addition, the frequency of patients with impaired DLCO at baseline did not change significantly after the first (37.5 vs. 45%, p = 0.581) and the second treatments (27.3 vs. 27.3%, p = 1.000).

Percent change in DLCO values (Fig. 1) after the first treatment did not significantly correlate with radiation dose (r = 0.221, p = 0.170), lung shunt fraction (r = − 0.171, p = 0.292), or lung exposure dose (r = − 0.043, p = 0.792). Similarly, these three parameters did not also correlate with DLCO change after the second treatment (r = − 0.357, p = 0.282; r = 0.170, 0.617; and r = − 0.029, p = 0.933, respectively).

None of the patients developed radiation pneumonitis during follow-up.

SIRT is an established treatment modality for chemoresistant, unresectable, primary, or metastatic hepatic malignancies. During the development phase of this therapeutic modality, tumor-induced arteriovenous pulmonary shunting and possible occurrence of RP emerged as potential limitations of the technique [2, 4, 24–26]. Post-treatment development of interstitial pneumonia and presence of micro-spheres in the lung biopsy were reported by Lin et al. in 1994 in a hepatocellular carcinoma patient who was treated with resin micro-spheres and who had a pulmonary shunt of 17% [24]. After that, Leung et al. reported the death of three out of five patients developing RP after SIRT [2]. While Salem et al. observed no clinical radiation pneumonitis in a total of 58 patients receiving higher-than-recommended doses (> 30 Gy) [4], Dobrocky et al. published a case report where an asymptomatic patient had radiation pneumonitis within 3 months following treatment as confirmed both by imaging modalities and lung biopsy [26].

An association between deterioration of DLCO and radiation pneumonitis has been reported by many authors. Guerra et al. investigated the possible association between the extent of change in DLCO after external radiotherapy and radiation pneumonitis in 140 patients, based on the hypothesis that patients with symptomatic radiation pneumonitis would experience a significant decrease in pulmonary function that could be quantified using DLCO [27]. These investigators found that patients who experienced higher percent reductions in DLCO were more likely to have high-grade radiation pneumonitis. The main difference between that study and ours lies in the type of radiation exposure, which was internally administered in the latter.

The incidence of radiation pneumonitis may differ between internal vs. external beam radiotherapy as a result of exposure differences. In 46 patients undergoing endobronchial brachytherapy due to malignant airway obstruction, radiation pneumonitis occurred only in those who received external beam therapy [28]. Among 80 subjects who received SIRT-alone, only five developed radiation pneumonitis [2]. In patients receiving external radiotherapy, lung injury has also been reported in non-RT areas, although of lesser severity, and CD4 + lymphocytic alveolitis similar to hypersensitivity pneumonia was described in both lungs, regardless of radiation exposure [29].

Lungs are one of the most radiation sensitive tissues in the body, and among the anatomical components of the lung, alveolar capillary complex exhibits highest sensitivity [30]. Within days to weeks after radiation exposure, initial cytokine release is triggered. Generally, cytokine release occurs within 2 weeks and has no associated symptoms. Second phase of cytokine release starts 6–8 weeks after radiation which is associated with hypoxemia and lung hypoperfusion [31, 32]. In our study, timing of DLCO coincided with this phase to detect the inflammation. The reduction in DLCO is thought to reflect a limited reserve of gas exchange resulting from the potential toxicity of radiotherapy. Although the reduction in DLCO is generally considered to be subclinical, it may provide a sensitive marker of chemotherapy-induced lung injury [12]. In our study, absence of a significant reduction in DLCO may be accounted for by the lower degree of direct exposure to radiation as compared to external beam radiation, leading to minimal injury.

Since radiation pneumonitis was not detected neither clinically nor radiologically during in the follow-up period in our study, we can affirm that the degree of lung injury was not severe in our cohort. However, a comparison with previous cohorts, and particularly with that of Salem et al., shows significantly lower level of radiation exposure in our patients [4], possibly explaining the lack of a statistically significant association between treatment dose, radiation exposure, and the change in DLCO before and after treatment. However, it should be noted that our study was not designed to identify the lowest dose level associated with the development of radiation pneumonitis following SIRT. Nevertheless, our results may give an idea regarding the safe dose range for SIRT.

In this study, no patients who were routinely followed up radiologically with PET/CT imaging at 75 days after SIRT and clinically for 6 months developed any signs or symptoms consistent with radiation pneumonitis. In patients who received a repeated treatment, PET/CT images obtained approximately 150 days after the first treatment also showed no pathology. As pointed out by Dobrocky et al. who used comparable treatment doses, early identification of asymptomatic radiation pneumonitis patients may bear clinical significance [26]. As recommended by Sangro et al., detection of the decline in DLCO values may prove to have clinical utility in these patients, particularly when the high radiation exposure associated with CT scans is taken into consideration [11].

In some patients with hepatic malignancy, repeated doses of SIRT may be required, increasing the likelihood of radiation pneumonitis. About one fourth of the cases were retreated in this study. Furthermore, our study did not show a statistically significant change in DLCO values in association with the SIRT procedure in the second treatment session.

Our study had some significant limitations. Ideally, this study should be repeated with a larger sample size with true lung volumes. In our study, lung mass was not assumed to be 1 kg based on International Commission on Radiological Protection (ICRP). Recent prospective autopsy studies by Molina et al. showed that there was no significant relationship between lung weights, either individually or combined, with body length, body weight, or BMI for healthy adult men (n = 232) and women (n = 102) [20, 21]. These authors also found no significant difference between the lung weights of underweight, normal weight, overweight, or obese individuals for men and women. The authors, therefore, proposed the mean lung weights to be 395 and 445 g for the left and right lungs (total 840 g), respectively, for men and 299 and 340 g (total 639 g), respectively, for women. Even though this approach lead to higher calculations of absorbed lung doses, we thought it will be more logical. Finally, the results could be different in another study population with higher lung shunt fractions, e.g., in patients with hepatocellular carcinoma with major vascular invasion.

Recently, Sancho et al. showed that imaging with ^99mTc-MAA is essential in SIRT workup in their retrospective analysis of 532 consecutive patients [33]. However, in contrast to the study by Jha et al., some researchers have reported that planar lung shunt fractions were overestimated with ^99mTc-MAA scans [34–39]. There were several explanations for this result. First, there was a correlation between scan quality and lung shunt fraction, suggesting that low scan quality leads to overestimation [40]. Second, according to O’Doherty et al., the scatter correction should be used on pretreatment ^99mTc-MAA scans in order to more accurately assess the lung shunting percentage before therapy [41]. Lastly, tracer degradation leads to overestimation of lung shunt fraction [42]. Since no cases developed RP in this study and DLCO values did not change significantly after the SIRT procedure, we can predict that a possible overestimation of the absorbed lung dose will not change our results. Larger studies with comparable protocols (planar vs. SPECT/CT images, ^99mTc-MAA scans vs. ⁹⁰Y post-radioembolization images, ^99mTc-MAA scans vs. ⁹⁰Y-microsphere PET/CT images, etc.) are required to better delineate DLCO changes after SIRT, especially in patients with higher lung shunt fractions.

The association between the alterations in DLCO and the severity of potential radiation pneumonitis developing after SIRT may prove to be clinically important. Since no cases developed radiation pneumonitis in this study, a cutoff value for the decline in DLCO could not be defined. If the assumed relationship between the percent change in DLCO and severity of radiation pneumonitis proves to be valid, then it may be possible to identify potential candidates for radiation pneumonitis using DLCO monitoring.

Findings of this study do not suggest a significant change in DLCO values in association with the SIRT procedure, either at the first or the second treatment sessions. Further large studies possibly with different protocols are warranted to better delineate DLCO changes after SIRT in a larger spectrum of patients.