Clinical investigation
Risk for hemorrhage during the 2-year latency period following gamma knife radiosurgery for arteriovenous malformations

https://doi.org/10.1016/S0360-3016(00)01432-2Get rights and content

Abstract

Purpose: Radiosurgery does not immediately obliterate an arteriovenous malformation (AVM), and the risk for hemorrhage still persists until the AVM is occluded. There is controversy about whether this risk is altered after as compared to before radiosurgery. The aim of this paper is to study this topic further and to suggest a model to predict the risk for posttreatment hemorrhage.

Methods and Materials: The incidence of hemorrhages within the first 24 months following Gamma Knife radiosurgery was studied retrospectively among 1593 AVM patients, and was related to patient, AVM, and treatment parameters.

Results: Fifty-six patients experienced a hemorrhage in the latency period, representing an average annual incidence of 1.8%. The incidence of posttreatment hemorrhage was related to the patient’s age, AVM volume, minimum dose, and average dose delivered to the AVM nidus. Based on these observations, an equation was defined that could quantify the probability for a posttreatment hemorrhage to occur.

Conclusion: A model that can predict the probability for a hemorrhage within the first 24 months after radiosurgery is presented. The risk is higher for larger AVMs and for older patients, and it is lower when higher doses of radiation are used.

Introduction

Gamma knife radiosurgery (GKRS) is a commonly used treatment in the management of arteriovenous malformation (AVM). It carries, however, a major drawback: the risk for an AVM rupture still persists during the time period between the treatment and its final result. It is currently not clear if this risk differs from the natural course of the disease or not. Scientific publications have concluded that the risk for a posttreatment hemorrhage is decreased 1, 2, unchanged 3, 4, or even increased 5, 6 as compared to the natural course of the disease. This also reflects the current different opinions among neurosurgeons worldwide. How can the conclusions reached be so different? Most probably, it reflects the different opinions of the magnitude of risk for hemorrhage an untreated AVM carries, which has, for selective groups, been reported to be as high as 33% (7) and as low as 1% (8) per year. Most authors, however, report the annual risk to be between 2 and 6% 7, 9, 10, 11, 12, 13, 14 .

The lack of a generally accepted nominator to which the risk for posttreatment hemorrhage should be compared has contributed to a situation where different well-known and prestigious experts in the field of AVM management have concluded that radiosurgery should or should not be recommended for the treatment of AVMs. Naturally, this puzzles not only physicians who refer AVM patients for treatment, but also the patients, who run the risk of receiving different recommendations as to how their AVMs should be treated from different reputable physicians. Thus, the situation in this respect is suboptimal, and there is a need for a consensus. Most likely, this cannot be reached until a consensus is reached in regard to the natural course of AVMs.

From a patient perspective, however, it seems to be more important to know how high the risk is for a posttreatment hemorrhage rather than to know if the risk has been affected or not, as nontreatment is seldom an option for AVMs, taking the high cumulative risk for hemorrhage into consideration (13). Thus, if the initial question is rephrased from “does radiosurgery change the risk for hemorrhage in the latency period” to “how high is the risk for hemorrhage in the latency period” a consensus may be reached without the need to agree on what is the natural course of the disease.

It is important to differentiate between two different types of posttreatment hemorrhages. If a hemorrhage occurs within the latency interval between the radiosurgical treatment and its final result, then the risk for hemorrhage is a part of the management risk following the radiosurgical treatment. If a hemorrhage occurs after the final result has been reached, it is due to the risk of having a persistent AVM, as this patient should have been offered an additional treatment. This hemorrhage cannot be denoted as a management complication to the radiosurgical procedure, but to the failure not to treat the malformation again, just like a hemorrhage from an AVM embolized 1 year earlier cannot be ascribed to the embolization itself.

We know from earlier studies that the latency time between the radiosurgical treatment and obliteration is decreased when the average dose of radiation to the AVM nidus (Dave) is increased 2, 15. In addition, there is a relation between the minimum dose (Dmin) and the probability for obliteration (16). Based on these findings, we decided to investigate if the risk for hemorrhage is dose dependent, and if other factors can be related to the incidence of posttreatment hemorrhages. Finally, the aim was to suggest a model with the power to predict accurately the probability for a posttreatment hemorrhage to occur within the first 24 months after the treatment.

Section snippets

Methods and materials

All AVM patients treated with GKRS at the Karolinska Hospital during the period 1970–1995 (1259 patients) or at the University of Virginia during the period 1989–1990 (317 patients) were included in the study. The information for the University of Virginia patients was collected from a previously published paper (2). Eleven patients had multiple AVMs, but this was considered being 28 patients with one AVM each. Thus, 1593 patients were eligible for and included in the study.

The time at risk for

Results

The number of patients suffering from a hemorrhage within the first 2 years after GKRS was found to be 56, equaling an average annual incidence of posttreatment hemorrhage of 1.8%. Of those, 25 (45%) occurred within the first 6 months after the treatment.

Both Dmin (p = 0.0002), and Dave(p = 0.0003) were related to the incidence of posttreatment hemorrhage, while Dmax was not (p = 0.94). For both Dmin and Dave, the higher the dose, the lower the incidence of posttreatment hemorrhage. As Dmin and

Discussion

The present study does not differ from most studies estimating the risk for an AVM to hemorrhage in one important respect: the accuracy of the calculated risk for hemorrhage stands or falls with the accuracy of the clinical information the calculations are based on. Naturally, if significantly more hemorrhages have occurred than are reported, a too low risk for hemorrhage will be estimated. If so, the risk for hemorrhage without treatment, corresponding to Dmin = 0 in Fig. 1, will be

Acknowledgements

We wish to thank Prof. Ladislau Steiner, for previously sharing the data of the AVM patients treated at the University of Virginia from 1989 to 1990, which was published earlier (2), and was again used in this paper.

References (27)

  • R.J Brown et al.

    Unruptured intracranial aneurysms and arteriovenous malformationsFrequency of intracranial hemorrhage and relationship of lesions

    J Neurosurg

    (1990)
  • P.M Crawford et al.

    Arteriovenous malformations of the brainnatural history in unoperated patients

    J Neurol Neurosurg Psychiatry

    (1986)
  • C Davis et al.

    The management of cerebral arteriovenous malformations

    Acta Neurochir (Wien)

    (1985)
  • Cited by (0)

    View full text