Indirect cell death and the LQ model in SBRT and SRS

Abstract

High-dose hypofractionated SBRT and SRS indirectly kills substantial fractions of tumor cells via causing vascular damage. The LQ formula may work well for certain clinical cases of SBRT and SRS when the indirect/additional tumor cell death secondary to vascular damage is small. However, when the indirect cell death is extensive, the LQ model will underestimate the clinical outcome of SBRT and SRS.

Despite the increasing use of stereotactic body radiation therapy (SBRT) and stereotactic radiosurgery (SRS) for a variety of cancers in recent years, the underlying biological principles of this unconventional high-dose per fraction radiotherapy modalities have not yet been clearly elucidated. Some investigators concluded that the cell death in tumors after high-dose irradiation is due entirely to DNA-double strand breaks and that the radiobiology concepts for the conventional multi-fractionated radiotherapy such as the 5 Rs and the LQ model are sufficient to explain the excellent clinical result of SBRT and SRS [1-3]. On the contrary, in light of the increasing evidence that indirect cell death due to vascular injury and perhaps immune response are involved in the response of tumors to high-dose per fraction irradiation, other investigators concluded that the LQ model is invalid for SBRT and SRS [4-6]. Given the recent increasing trend to hypo fractionate the radiation exposures in clinical SBRT and SRS, it is imperative to critically asses the validity of the LQ model. In this letter, we propose a hypothesis regarding the validity of the LQ model in treating cancers with SBRT and SRS.

The survival and proliferation of cancer cells in tumors are closely related to the functions of various components of the intratumor microenvironment, in particular blood vessels. Park et al. [7] concluded that irradiation of tumors with doses higher than approximately 10 Gy induces severe vascular damage and that the resultant breakdown of the intratumor microenvironment leads to secondary or additional tumor cell death. Over the last decades, a host of other groups have made similar observations regarding the indirect tumor cell death due to vascular injury after high-dose irradiation. Lasnitzkin [8] reported in 1947 that, in mouse adenocarcinoma, as much as two-thirds of the total cell death following 2000-3000 rads (20-30 Gy) in a single dose was due to vascular damage. In 1973, Nicolas and McNally [9] reported that when RIB5 rat fibrosarcoma was irradiated with 2000 rads in a single dose, the extent of tumor cell death estimated from the length of tumor growth delay was 10 times greater than the cell death determined soon after irradiation by the in vivo-in vitro excision assay method. The authors attributed the extended tumor growth delay to additional tumor cell death due to vascular damage and ensuing decline in supply of nutrients to the tumor cells. In the 1970s, Clement and his colleagues [10] found that a single exposure to 1000 or 2000 rads caused severe vascular damage yielding marked indirect cell death in the Walker rat tumor model. More recently, similar indirect cell death due to vascular injury after high-dose irradiation was observed in human melanoma xenografts [11], HT-180 human sarcoma xenografts [12], FSaII fibrosarcoma of mice [12, 13] and MCA-129 fibrosarcoma xenografts [14]. In agreement with these observations on the radiation-induced vascular damage, doses higher than 8-10 Gy caused massive apoptosis of endothelial cells via activation of the acid sphingomyelinase/ceramide pathway [15]. It was thus concluded that the indirect tumor cell death secondary to the death of endothelial cells regulated tumor response to radiotherapy. Indirect tumor cell death secondary to vascular damage after high-dose irradiation has also been observed in human patients. As much as 19-33% of cell death in human cranial lesions following SRS could be attributed to ischemic injury [16]. In human intracranial neoplasms, the microvascular endothelial cells were found to be the primary target of a single high-dose radiosurgery [17]. Based on these unequivocal clinical and pre-clinical observations of additional tumor cell death following high-dose irradiation, a hypothetical radiation survival curve of tumor cells in vivo has been proposed, as shown in Figure 1a [7, 12, 13]. It is demonstrated that, by virtue of the additional cell death due to vascular damage, the tumor cell survival curve bends downward with increasing slope as the radiation dose exceeds higher than about 10 Gy in a single exposure. A similar radiation survival curve of tumor cells reflecting additional cell death in vivo was also reported by other investigators [4].

Figure. 1.

a) the radiation survival curve of tumor cells in vivo — the dotted line indicates the cell death if the cell death in vivo is only due to DNA breakage; b) the LQ model cell survival curve. The dotted line indicates the cell death in vitro due to DNA breakage; c) the cell survival curves based on the LQ formula (A) is shown with the cell survival curves in vivo representing the cell death due to moderate vascular injury (B) and extensive vascular injury (C).

Figure 1b shows a typical radiation survival curve of tumor cells calculated with the LQ model. It is shown that the slope of the survival curve continuously increases (bends downward) with increasing radiation dose even though the survival curve is supposed to be linear with a constant slope (Do) at higher doses, as indicated with the dotted line. Accordingly, the LQ model has historically been predicted to overestimate the radiation-induced cell killing or tumor damage [18]. However, the LQ model has often been reported to underestimate the efficacy of SBRT and SRS treatments [4-6,13]. Given that the LQ model does not include the potential impact of high-dose irradiation on many important constituents of tumor biology including vascular and immune response, it is not surprising that the LQ model could underestimate the total tumor cell killing induced by SBRT and SRS. Furthermore, the LQ model does not reflect the potential reoxygenation of hypoxic cells following an initial radiation exposure leading to increased tumor radiosensitivity [10, 19]. It is of note that the survival curve of tumor cells in vivo (Figure.1a) and the LQ model predicted cell survival curve (Figure 1b) are strikingly similar in that they both deviate downward from the original linear slope (dotted lines) proportional to increasing radiation dose. Taken together, these observations suggested that what was thought to be an inherent flaw of the LQ model in overestimating the cell death at high radiation doses is actually nullified by the reality of additional tumor cell death caused by vascular dysfunction in vivo. In Figure 1c, the cell survival estimated based on the LQ model (curve A) is shown together with two hypothetical cell survival curves representing different clinical cases with different magnitude of indirect cell death. The LQ model will underestimate the total cell death if indirect cell death by irradiation is extensive as indicated by curve C. On the other hand, in clinical cases when indirect cell death is relatively moderate (curve B), the cell death predicted by the LQ model (curve A) may be comparable with the actual total cell death. We may hypothesize that the LQ model may work with varying accuracy for predicting the outcome of SBRT and SRS regimens depending on the extent of the additional cell death through indirect mechanisms. Interestingly, the clinical data for early stage non-small cell lung cancer were in good agreement with expectations based on the LQ model calculation [3]. It is conceivable that the predicted cell death by the LQ model in lung tumors was coincidently similar to the total tumor cell death due to a combination of direct cell death caused by DNA damage and indirect cell death induced via vascular damage. Brown [20] recommended to “Just stick with straight LQ” in treating cancer with SBRT. Note that, as Kirkpatrick stated, we should not accept the LQ model without any question even if it matches reasonably well with the observed data [4].

It may be further suggested that the LQ model may work well for occasional clinical cases of SBRT and SRS not because tumor cells are killed only through the classical radiobiological principles such as the 5 Rs but because substantial fractions of tumor cells are killed by mechanisms other than the 5 Rs. It seems more and more certain that “New radiobiology” beyond direct DNA damage plays an important role in the response of tumors to SBRT and SRS.