Radiation-induced cavernous hemangiomas of the brain

A late effect predominantly in children

Authors

  • Stefan Heckl M.D.,

    Corresponding author
    1. Department of Oncological Diagnostics and Therapy, German Cancer Research Center, Heidelberg, Germany
    • German Cancer Research Center Heidelberg, Department of Oncological Diagnostics and Therapy, Im Neuenheimer Feld 280, 69120 Heidelberg, Germany
    Search for more papers by this author
    • Fax: 011–0049-6221-42-2514

  • Alfred Aschoff M.D.,

    1. Department of Neurosurgery, University of Heidelberg, Heidelberg, Germany
    Search for more papers by this author
  • Stefan Kunze M.D.

    1. Department of Neurosurgery, University of Heidelberg, Heidelberg, Germany
    Search for more papers by this author

Abstract

INTRODUCTION

The induction of cavernomas as a consequence of brain irradiation was first suspected in 1994 and has been controversial since that time.

METHODS

Between 1986 and 2000, 189 cerebral cavernomas were diagnosed in the Neurosurgical Department of the University of Heidelberg; of those patients, 5 had received prior radiation therapy. The ages of these 5 patients were compared with those of the 184 others with naturally occuring cavernomas. In an examination of 40 patients with cavernomas occurring after radiation (the 5 mentioned above, plus 35 from the literature) the age distribution was investigated, and a possible relationship between radiation dosage and latency interval to diagnosis of cavernoma was examined.

RESULTS

Almost one in four of the patients under 15 years of age diagnosed with a cerebral cavernoma in the Neurosurgical Department of the University of Heidelberg had received prior radiation. In 40 patients with cavernomas and prior radiation (5 from Heidelberg, 35 from the literature), there was a clear accumulation in the age group of 10–19 years (50%). Most of those patients had received radiation in the first 10 years of life. The accumulation of cavernomas after radiation in childhood could not be explained by a greater frequency of radiation exposure in children compared to adults. In children up to 10 years of age at the time of radiation therapy, a dose of 3000 cGy and higher was followed by a shorter latency interval to incidence of cavernoma (P = 0.0018). In patients older than 10 years at the time of radiation, postradiation cavernomas only occurred when dosage was 3000 cGy or greater.

CONCLUSION

These results indicate a correlation between radiation and cavernoma, particularly in children under 10 years of age at the time of radiation therapy. In adults, cavernomas after radiation rarely occur, and then only after higher radiation dosages (3000 cGy or more). Cancer 2002;94:3285–91. © 2002 American Cancer Society.

DOI 10.1002/cncr.10596

There are clear indications that gliomas and meningiomas can result from prior irradiation.1 It has also been suspected that radiation could lead to the development of cavernomas, but so far this has not been clarified. Ciricillo et al. in 19942 first described seven patients with intracerebral cavernomas as a possible consequence of radiation. Twenty eight further cases have since been described in the literature.2–17

Between 1986 and 2000, 189 cerebral cavernomas were diagnosed at the Neurosurgical Department of the University of Heidelberg. Five had received prior irradiation to the brain. Until now it was uncertain whether there was a causal relationship or whether this was merely an incidental finding. By means of statistical analysis we have attempted to clarify this issue.

PATIENTS AND METHODS

In the Neurosurgical Department of the University of Heidelberg, 189 patients with cavernomas were diagnosed by MRI, 132 of whom subsequently underwent operations. Of these 189, 5 had received prior radiation (latency interval to diagnosis of cavernoma: 1.16 to 7 years; Fig. 1A). These five patients were three children aged 5 years and two adults of 27 and 29 years at the time of radiation treatment. The ages of these 5 patients were compared to those of the 184 patients with naturally occurring cavernomas. By means of the Fisher exact test, we examined whether cavernomas after radiation occurred more frequently in patients under 15 years of age at time of diagnosis.

Figure 1.

A and B. Age distribution at the time of diagnosis of cavernoma. A: One hundred eighty nine patients (Neurosurgical Department, University of Heidelberg, 1986-2000). Solid: with prior radiation; clear: without prior radiation. B: Forty patients with prior radiation (5 Heidelberg, 35 from the literature).2–17 C: Age distribution at the time of radiation. Solid: 4208 patients receiving brain irradiation of whom 2039 had known survival times; striped: 365 patients with a survival time > 1.16 years. Radio-Oncological Department, University of Heidelberg, 1982-2000.

In two children aged five years at the time of radiation, prophylactic whole brain radiation with additional chemotherapy had been performed due to acute lymphoblastic leukemia (ALL). Another child aged five years with rhabdomyosarcoma of the sphenoid sinus and two adult patients with brain tumors had been operated and irradiated afterwards (with additional chemotherapy in the case of rhabdomyosarcoma).

The exact radiation dosage in the region of the subsequently occurring cavernoma was examined by means of isodose-lines (500 to 3200 cGy, Fig. 2).

Figure 2.

Thirty three year old man: cavernoma right precentral, six years after radiation (56 Gy) of an oligoastrocytoma, left frontal (World Health Organization Grade II-III).

The age distribution of 40 patients with cavernomas after radiation was also examined (5 cases at the University of Heidelberg and 35 from the literature;2–17 Fig. 1B).

The accumulation of cavernomas after brain irradiation in children could simply be explained by a greater frequency of radiation therapy in children compared to adults. For this reason, the age distribution of all patients receiving brain irradiation between 1982 and 2000 at the University of Heidelberg was examined (Fig. 1C). There were 143 patients aged 10 years or younger, with the remaining 4065 patients older than 10 years. Of the 143 patients aged 10 years or younger, 3 showed cavernomas after radiation. Of the 4065 patients older than 10 years, only 2 showed cavernomas. By means of the Fisher exact test, the complete group of 4208 patients was tested to see whether cavernomas after radiation occurred at a statistically significant higher rate in children irradiated at 10 years of age or under than in patients above this age.

To minimize a possible selection bias due to longer survival times in children compared to adults causing an apparent higher relative incidence of cavernomas, only those patients were taken into consideration in the further analysis for whom, firstly, the survival time was known (2039 of 4208) and, secondly, a minimum latency interval requirement of 1.17 years (approximately 1 year and 2 months) was fulfilled, which reduced the total patient number to 365 (Fig. 1 C and Fig. 3). This latency interval was chosen because 1.16 years was the shortest latency interval before the observed incidence of cavernoma. The above filtering process was also conducted using values of three and five years for the minimum latency interval (five years was the second shortest latency time to cavernoma in the Heidelberg patient group, three years was taken as the approximate mean value of the first and second shortest latencies).

Figure 3.

Three hundred sixty five patients (survival time > 1.16 years, Radio-Oncological Department, University of Heidelberg, 1982-2000) grouped by age and irradiation dosage. Occurrence of cavernoma by group.

Furthermore, the histories of the 40 patients who had developed cavernomas after radiation therapy were examined to ascertain whether cavernomas were observed more frequently in children due to more frequent follow-up magnetic resonance imaging (MRI) or due to the signs and symptoms presenting earlier or more clearly.

Using the two-sided Wilcoxon test, we investigated whether, in children aged 10 years or younger (24 cases; 3 from Heidelberg, 21 from the literature),2, 4–6, 8–11, 15, 17 the development of cavernomas was related to the total dosage using the threshold value 3000 cGy to further subdivide the under 10 years age group and, furthermore, whether the interval between radiation and the diagnosis of cavernoma was dependent on the total dosage (Fig. 4). Finally, we examined whether a total dosage less than 3000 cGy led to the occurrence of cavernomas after radiation in patients older than 10 years at the time of treatment (2 from Heidelberg, 13 from the literature,2, 3, 7, 9, 11–17 Fig. 4).

Figure 4.

Graph of age at the time of radiation therapy versus latency of cavernoma as a function of the total irradiation dose for 39 patients (5 from Heidelberg, 34 from the literature).2–4, 6–17 Solid: ≥ 30 Gy; clear: < 30 Gy.

RESULTS

A connection between cavernomas and radiation has been proposed in previous papers,2–17 but to date there have been no conclusive statistical results. The current results, however, show that the assumption of an accidental occurrence of cavernoma after radiation is no longer justified, at least in the case of children.

In all, 189 cerebral cavernomas have been diagnosed by the Neurosurgical Department of the University of Heidelberg using MRI (132 of whom underwent surgery, Fig. 1A). Of the 189 patients, 13 were less than 15 years of age. Of these 13 children, 3 had received prior radiation treatment (3 out of 13, 23%). (Fig. 1A)

This implies that almost one in four of the patients under 15 years of age diagnosed with cerebral cavernoma in the Neurosurgical Department of Heidelberg had received prior radiation to the brain.

Of the remaining 176 patients who were 15 years or older at the time of cavernoma diagnosis only 2 had been previously subjected to radiation (2 out of 176, 1.1%). This would suggest that cavernomas in patients previously subjected to radiation as children occur at a significantly higher rate (two-sided Fisher exact test: P = 0.0042 Fig. 1A).

Of the 184 patients with naturally occurring cavernomas diagnosed in the Neurosurgical Department of the University of Heidelberg, the largest group was aged 30–45 years (35% Fig. 1A).

Most of the 40 patients with cavernomas after radiation (5 from Heidelberg, 35 from the literature) were aged between 10 and 19 years at the time of diagnosis (50%; Fig. 1B). This underscores the high susceptibility of children. One could attempt to explain the high number of children with cavernomas after radiation by a higher number of children receiving radiation treatment compared to adults; however, upon examination, the contrary becomes apparent (Fig. 1B vs. Fig. 1C). In the Radio-Oncological Department of the University of Heidelberg, far fewer children up to 10 years of age were irradiated (143 out of 4208, 3%) than patients older than ten years (4065 out of 4208, 97%). Nevertheless, the 40 patients showing cavernomas after radiation (5 from Heidelberg, 35 from the literature) were most often irradiated in the first decade of life (25 out of 40, 63%).

Thus, the suspected radiation induced cavernomas occurring between ages 10 and 19 years cannot be explained by a higher total number of brain irradiations in this group.

In Heidelberg, cavernomas after radiation were seen in 3 out of 143 children aged 10 years or younger (2%) whereas only 2 out of 4065 patients older than 10 years showed a cavernoma (0.05%). The natural occurrence of cavernomas varies between 0.02 and 0.53%.18

One could assume that children less than 10 years of age have a longer survival time than adults and thus have a longer period in which to develop a cavernoma. In this case, the survival time would be the most relevant factor in the development of a cavernoma and not the age of the child. In order to preclude this, only those patients whose survival time was known were included in the further analysis (2039 out of 4208). The shortest latency time for development of cavernoma observed in the Heidelberg group was 1.16 years; therefore, the period of 1.17 years was chosen as the minimum required survival time for inclusion in the further analysis, which excluded 1674 out of 2039 cases.

The total number of patients whose survival time was known and who fulfilled the minimum survival time requirement was 365 (Fig. 3). This patient collective was further divided into two groups, one aged 10 years or less at the time of radiation therapy, the other aged greater than 10 years.

Twenty one out of 365 were aged 10 years or less at the time of radiation therapy, 3 of whom later developed a cavernoma. The remaining 344 were older than 10 years at the time of therapy, 2 of whom later developed a cavernoma. The Fisher exact test was used to investigate whether cavernomas were more frequent in children aged ten years or less when the survival time cut-off criterion (1.17 years) was utilized (two-sided Fisher exact test: 0.0015; Fig. 3). This would imply a factorial difference of approximately 24.

Even if the survival time cut-off point is extended to three or five years (five years was the next shortest latency interval, three years was the approximate mean of the two shortest latencies), the incidences in the the two age groups remained proportionately similar. When the survival time cut-off was set at 3 years, there were 3 cavernomas resulting in 8 children aged 10 years or less at the time of radiation therapy, compared to 2 cavernomas resulting in 68 patients aged greater than 10 years (P = 0.0072). In the case of a 5 year cut-off point, 3 cavernomas resulted in 5 children aged 10 years or less at the time of radiation therapy, compared to 1 cavernoma resulting in 21 patients aged greater than 10 years (P = 0.0144).

As illustrated in Figure 3 (survival time cut-off 1.17 years), these two age specific groups (threshold age 10 years at the time of radiation therapy) were each divided further into two dosage specific groups using the threshold value 3000 cGy. The Fisher exact test was used to compare the incidence of cavernoma after radiation in the respective equal dosage groups, and a significant difference was found between the groups receiving 3000 cGy or more (P = 0.0055; Fig. 3).

In the next step, the length of the latency period between the age at radiation therapy and the age at the appearance of cavernoma was investigated. A dependency on total radiation dosage was also examined.

In children up to and including 10 years of age (3 from Heidelberg, 21 from the literature),2, 4–6, 8–11, 15, 17 the length of the latency period clearly depended on the total radiation dosage. The higher the dosage (≥ 30 Gy), the shorter the latency period (Fig. 4).

The median time between the age at radiation therapy and the age at diagnosis of cavernoma was 10.5 years in the case of a total dosage under 3000 cGy and 4.75 years when the applied dosage was 3000 cGy or higher (two-sided Wilcoxon test, P = 0.0018; Fig. 4). For patients older than ten years of age at the time of radiation therapy (2 from Heidelberg, 13 from the literature),2, 3, 7, 9, 11–17 cavernomas after radiation occurred in the vast majority of cases only when the applied total dosage was 3000 cGy or higher (Fig. 4).

It is conceivable that younger patients who develop a cavernoma have more distinct or faster developing symptoms and are therefore recognized earlier or more easily than in the case of adult patients. For this reason, all 40 patients with cavernoma after radiation therapy (35 from the literature, 5 from Heidelberg) were regrouped by signs and symptoms which led to diagnosis (Table 1). However, no particular differences in signs or symptoms could be observed between children and adults. Approximately 50% of patients in both groups presented with clinical symptoms; the other 50% were symptom free and were diagnosed in MRI followup examinations. There was no difference in the location of naturally occurring cavernomas and those occurring after radiation.

Table 1. Presenting Symptoms of Patients with Cavernomas After Radiation
Age< 20 years (n = 24)≥ 20 years (n = 16)
  1. MRI: magnetic resonance imaging.

No symptoms (regular MRI controls)138
Epileptic seizures12
Seizures, focal deficits30
Seizures, focal deficits (hemorrhage)11
Focal deficits02
Focal deficits (hemorrhage)20
Headache, vomiting22
Headache, vomiting, focal deficits10
Headache, focal deficits (hemorrhage)11

The patient collective included various diagnoses and a resulting variety of treatment schedules which in turn made it difficult to specify a single follow-up protocol for the entire patient group.

DISCUSSION

As a consequence of this situation and the large catchment area from which the 4208 patients (Heidelberg) derived, it cannot be claimed with total confidence that absolutely all cases of cavernoma after irradiation were captured in the current analysis. It is arguable that younger patients who have undergone radiation therapy are likelier to receive closer surveillance testing of the brain than older patients who have not undergone radiation therapy; however, it should be noted that almost exactly half of each age specific group (Table 1) were diagnosed on the initial basis of symptoms shown (5 cases, from Heidelberg 35 from the literature) which were then confirmed with imaging techniques. The remaining half were symptom free and were diagnosed in the course of MRI follow-up examinations. Still, in the investigated group of 365 patients filtered for known minimal survival time of at least 1.17 years who underwent regular MRI follow-up examinations in Heidelberg clinics, there remained at least one order of magnitude between the frequencies of cavernomas after irradiation in the age groups above and below 10 years of age at time of radiation therapy.

It is unclear why irradiation of the brain can lead to the development of a cavernoma. It is not necessarily the case that cavernomas result directly from radiation. It could be assumed that children with ALL— the most frequent reason for brain irradiation—are inherently prone to the development of cavernomas. There are, however, no indications of this in the literature.

One could also postulate that the growth of a preexisting tiny cavernoma, not visible in MRI (especially computed tomography) prior to radiation, could have been induced, and the de novo appearance would merely have been simulated.

But why should this appear in a significantly higher rate in children than in adults? Mechanisms of induction of cavernoma by radiation remain shrouded in mystery.

An involvement of the release of vascular endothelial growth factor (vEGF) by radiation would be likely to result in the induction of angiogenesis.19 Radiation first results in a narrowing of the vascular vessel lumen due to adventitial fibrosis and endothelial edema. This leads to ischemia and micro-infarctions followed by reactive neoangiogenesis due to the increased release of hypoxy-inducible-factor-1 (HIF-1), which in turn induces VEGF.20 The evolution of some cavernomas distant from the radiation port could be explained by the long distance effect of VEGF. Vascular endothelial growth factor and HIF-1 have been found to be expressed most in the very young.20, 21

This most probably plays a part in the explanation of the particular frequency of radiation induced cavernomas in children.

Ancillary