Should Cancer Survivors Fear Radiation-Induced Sarcomas?

Purpose/Results. Ionizing radiation is carcinogenic and the induction of a second malignancy is a serious potential long-term complication of radiotherapy. The incidence of radiation-induced sarcomas was evaluated from many large epidemiological surveys of long-term cancer survivors reported in the literature over the past 30 years and only one case was found for every 1000 patients irradiated. Discussion. Although greater numbers of cancer patients are receiving radical radiotherapy and surviving free of disease for longer intervals, cases of radiation-induced sarcomas are rare and should not deter patients from accepting radiotherapy as treatment for curable cancers. With improvements in the administration of radiotherapy over the past two decades which are resulting in less damage to bone and soft tissues, it is likely that fewer cases of this condition will be seen in the future. If these sarcomas are diagnosed early, long-term survival can be achieved with surgical excision and possibly re-irradiation, as occurs in other types of sarcomas.


Introduction
Advances in cancer treatments are producing increasing num bers of healthy long-term survivors and, as a consequence, the delayed com plications of anti-cancer therapies are assu m ing a greater signi® cance. Radiation has been recognized both as a therapeutic agent capable of curing localized m alignancies and as a m utagen w hich may induce new cancers. It is just under 100 years since the ® rst cases of radiation-induced tum ours w ere recognized, in the ® rst decade following the discovery of X -rays by Ro È ntgen in 1895, w hen pioneer radiographers were reported with squamous cell carcinom as on exposed hands 1 and with leukaem ias. A soft tissue sarcom a induced by radiation was described in 1904 by Perthes 2 and, in the 1920s, bone sarcom as were noted follow ing irradiation for tuberculous arthritis and radium exposure of workers painting watch dials. 3,4 C ancer patients m ay develop second neoplasm s spontaneously or as a consequence of the sam e genetic or environm ental factors w hich led to the developm ent of their ® rst tum our, and the role of radiation can be dif® cult to establish since no speci® c histological or biochemical m arkers have yet been identi® ed. How ever, speci® c cases m ay be attributed to radiotherapy when they develop in norm al tissues that have been previously irradiated, follow ing a reasonable induction interval, particularly if the histology is unlike the usual tum ours of the region.
Sarcom as can be categorized as radiation induced if they m eet the following criteria, adapted from the original prerequisites established by Cahan et al.: 5 (1) the sarcom a m ust develop within the boundaries of a previously irradiated area; (2) a relatively long asym ptom atic latent period (at least 4 years) m ust have elapsed; (3) the sarcom a must have a different histology from the original lesion; (4) the sarcom a m ust be histologically con® rm ed.
Bone and soft tissue sarcom as induced by therapeutic irradiation are frequently advanced by the time they are diagnosed and are usually incurable. W ith current trends to use m egavoltage radiotherapy for organ conservation as an alternative to surgery, often in com bination with intensive chem otherapy protocols, in growing num bers of cancer patients, it is im portant to establish how com m only these cases occur and ® nd w ays to diagnose them early when they are am enable to the curative treatm ents that are applied for prim ary sarcom as. T he literature was reviewed to de® ne the incidence of cases of bone and soft tissue sarcom as which have 1357-714 X/97/010005± 11 Ó 1997 Journals Oxford Ltd developed within radiation ® elds of patients who w ere treated by radiotherapy and survived beyond a reasonable latent period to allow time for the developm ent of a radiation-induced sarcom a.

Radiation-ind uced sarcomas
A literature review of 344 sarcom as following irradiation undertaken by Robinson and colleagues 6 in 1988 found that the m edian latent interval was 11 years and the m edian survival was 12 m onths, with only 11% of patients alive after 5 years. M ost patients were diagnosed at an advanced stage with sarcom as that were high grade and usually unresectable, often developing m etastases and unresponsive to chem otherapy. Histologically, the largest group com prised osteosarcom as, followed by ® brosarcom as, m alignant ® brous histiocytomas, angiosarcom as, chondrosarcom as and others. T he com m onest prim ary w as breast cancer, then gynaecological cancers and retinoblastom a.
The range of radiation doses found to induce sarcom as is wide and tends to be higher for m egavoltage than orthovoltage (DX R) treatments. T hese m odalities have different photon abso rption patterns w hich have led to the phasing out of orthovoltage radiotherapy for the m ajority of cancer patients over the past 40 years. Bone tum ours and sarcom as in cutaneous and subcutaneous tissues are m ore comm on after orthovoltage therapy, 7 which delivers to bone approxim ately twice the dose that is absorbed by adjacent soft tissues. Som e reviews have noted a dose± response relationship for radiation doses in excess of 10 Gy, 8,9 below which very few cases of radiation-induced sarcom as have been reported. It has not been established whether children are at higher risks than adults, allow ing for the longer time interval for them to develop sarcomas. 10 Population-b ased studies of radiation-induced sarcom as have uniform ly found very low incidence rates. F igures reported from large cancer centres range from 0.03% and 0.38% of all 5-year survivors 8,11,12 to 0.27% of all 10-year survivors. 13 Between 1% and 3% of all sarcom as have been asso ciated with prior irradiation to the sarcoma site, and soft tissue sarcom as are three times as comm on as bone sarcom as. 7,14,15 A num ber of fam ilial conditions have been linked to m ultiple cancers including sarcom as, and it is likely that radiation will increase the incidence of sarcom as occurring in these patients. T his has been long app reciated in retinoblastom a patients, 16 where the genetic cases with bilateral tum ours have a high rate of sarcom as within the radiation port as w ell as at more peripheral sites. N euro® brom atosis 17 and the Li-Fraum eni syndrom e 18 are tw o other autosom al dom inant genetic conditions where sarcom as m ay develop in the absence of radiotherapy, and the sarcom a classi® ed as radiation induced when this is but a secondary factor.
Other factors which have been asso ciated with the developm ent of sarcom as in large num bers of case reports include excessive radiation dam age to skin, bone and soft tissue, and the contributory effects of chem otherapy. The m ajority of cases have been reported following orthovoltage radiotherapy, w here frequently higher daily dose fractions were given than is acceptable in current radiotherapy practice. Patients were often exposed to m ultiple courses of radiotherapy to the sarcom a site and total doses were com m only far in excess of what is now regarded as the radiation tolerance dose for the tissue. It is probable that som e chem otherapy regim ens stim ulate sarcom a induction, 19± 21 and in children at least, alkylating agents m ay double the risk. 9 For m any cancers the total doses of adjuvant chem otherapy have been reduced in recent years follow ing evidence that less intensive regim ens are equally effective.

Retinoblastomas
Retinoblastom a is a cancer affe cting young children which can be cured by radiotherapy. Inherited in around 40% of cases, genetic cases are usually bilateral and follow an autosom al dom inant pattern of inheritance. There is a strong association with sarcom as, which occur both within radiation portals and at other sites, frequently in patients w ho have not been given radiotherapy.
The largest study of 1603 long-term retinoblastoma survivors from N ew York and Boston hospitals treated from 1914 to 1984 by Eng and colleagues 22 found that 6.6% died from second tumours after a m edian follow-up of 17 years from radiotherapy, resulting in a cum ulative probability of death from second prim ary neoplasm s of 26% at 40 years following diagnosis for patients w ith bilateral disease. H owever, this report included cases other than sarcom as, and an unspeci® ed num ber occurred outside the ® eld of irradiation. Second tumours are not always fatal, and m any are cured by aggressive treatment. 23 Radiotherapy has a well-established role in the treatment of retinoblastom as, curing early cases with the preservation of useful vision. Bone and soft tissue sarcom as may develop irrespective of w hether the patient has received radiotherapy or not, and the recom m ended optim al radiation dose is now half that previously given. As a further bene® t, it has been noted that prophylactic retinal radiotherapy can signi® cantly reduce the incidence of contralateral retinoblastom as in fam ilial cases, using a radiotherapy technique where the exit dose passes through the clinically unaffe cted eye. 24 T ab le 1. Breast cancer: radiation-induced sarcom as following radiotherapy

B reast cancer
M any cancer centres and registries have reported radiation-induced sarcom as in long-term survivors of breast cancer treated with radiotherapy, 8,11,13,25± 33 and these are sum m arized in Table 1, which includes over 50 000 patients treated over the past 70 years with m edian follow -up of at least 5 years. T he overall incidence for radiation-induced sarcom as w as 0.10% (53/53 328). M ost bone sarcom as arose in the scapula, hum erus, clavicle or ribs and were osteosarcom as, and the majority were incurable. 34 Incidences were higher in series using orthovoltage radiotherapy than in later studies using m egavoltage irradiation and, in the recent reports, soft tissue sarcom as outnum bered bone sarcomas. T he highest incidence reported in a recent series was 1.10%, from 1382 consecutive autopsies on w om en dying w ith breast cancer (Roswell Park 1956± 1988, but this is not com parable to the other reports because of selection factors. 35 The large Swedish registry study of K arlsso n et al. 33 reviewed in detail all cases of soft tissue sarcom as in 13 490 wom en treated for breast cancer over a 20-year period and found the m ean annual incidence of soft tissue sarcom as w as doubled to 0.02% com pared to the norm al population. O nethird of their 18 cases had not received radiotherapy to the sarcom a site and, using a case-control analysis, half the radiation-induced sarcom as were found to have lym phoedem a and/or high radiation doses as predisposing factors. The very low risk of sarcom a developm ent, it was concluded, was likely to be even low er with the reduction in the dose and volume of breast irradiation and rarity of lym phoedema using the com bined techniques of radiotherapy and surgery now widely adopted. Current standard practice consists of tangential beam irradiation with com puter-assisted planning and it is uncom m on to irradiate the axilla or supraclavicular fossa. D ose fractionation protocols have evolved to ensure good cosmesis is a major objective and bone and soft tissue com plications are now rare, particularly at ® eld junctions, w hich were a com m on site for sarcom as to arise. 29,30,36 M any recent reports have described sarcom as arising in conserved breasts following the increased use of lum pectomy and breast irradiation as an alternative to m astectomy. T he potential risk of radiation-induced sarcom as has been a controversial issue, and three cases have been reported in over 3000 patients followed prospectively since 1973, from the M ilan C ancer Institute, the centre which reported the ® rst randomized trial in breast conservation. O nly one of the three cases was fatal. T he rarity of these cases was such that no change in the current policy of conservative therapy with radiotherapy for breast cancer was advocated. 32 Angiosarcom as of the breast or overlying skin are now the m ost com m on sarcom a following breast conservation treatment, 37 and have a better prognosis than the other types of sarcom a, with m ost cases cured by m astectom y, particularly if they are diagnosed early. M alignant ® brous histiocytomas and ® brosarcom as have also been reported. 38 M any case reports describe sarcom as which arose in areas of chronic radiation dam age from excessive doses of orthovoltage radiotherapy, at junctions of adjacent radiation ® elds or at hot spots when current methods of com puterized planning w ith tissue density corrections were not availab le. It is highly likely that such cases do not occur with m odern radiotherapy practice, w here a great deal of care is taken to avoid poor cosm etic results when breast conservation techniques for early cases of breast cancer are utilized as an alternative to m astectom y. T he com m on sites of bone sarcom as, the scapula, hum erus and clavicle, are usually om itted from current radiotherapy techniques, which do not routinely treat nodal regions.

H odgkin' s disease
Patients with Hodgkin' s disease are usually young and the vast m ajority are cured by radiotherapy and/or chemotherapy with few com plications. M O PP (m ustine, vincristine, procarbazine, prednisone) chem otherapy m ay induce acute nonlymphocytic leukaem ia in the ® rst decade of followup, but thereafter a growing num ber of patients are at risk of developing solid tum ours, a sm all proportion of w hich are sarcom as induced by their therapeutic irradiation. M any cancer centres have reviewed their long-term survivors to report increasing rates of second m alignancies with longer periods of follow-up, and collaborative groups have combined national and international cohorts of patients to assess the risks of sarcom a developm ent, as detailed in Table 2. In 15 studies w hich followed a total of 69 000 patients w ith Hodgkin' s disease treated since 1943, only 0.14% developed sarcom as following radiotherapy. 39± 53 How ever, som e reports do not provide suf® cient details to determ ine w hether all their cases would qualify as radiationinduced sarcom as. The m ost consistently reported solid tum ours found follow ing radiotherapy are non-H odgkin' s lym phom a, breast cancer and lung cancer.
Factors other than radiotherapy which are potentially related to the induction of these second m alignancies include chem otherapy, which w as given to the m ajority of patients, im m unological abnorm alities and a genetic predisposition, while other cases m ay be coincidental and found through the closer m edical surveillance that occurs in young cancer survivors.

Testicular tum ours
T esticular cancers, like H odgkin' s disease, generally affect young adults and have cure rates exceeding 90% , with sem inom as com m only treated by irradiation and teratom as usually receiving chem otherapy follow ing orchidectomy. M any cancer institutions, national cancer registries and international cooperative groups have review ed their long-term testicular cancer survivors and reported on second m alignancies, and the larger studies are detailed in T able 3. 10,39,54± 62 Some increases in the incidence of lung, gastrointestinal and genitourinary cancers have been found, but no study has found a signi® cant risk of bone sarcom as, and the relative risk for soft tissue sarcom as varies from 1.0 to 5.4. 59 O f the 50 000 patients listed in the table, followed for a m edian duration of 5± 15 years, 0.05% developed a subsequent sarcom a, and this ® gure includes som e cases w hich m ay not have received radiotherapy or developed the sarcom a in tissues outside the radiation ® eld.
M odi® cations to the treatment of testicular tum ours in recent years include signi® cant reductions in the dose and volum e of radiotherapy, the om ission of routine irradiation to the m ediastinum , and the increasing use of surveillance and chem otherapy without irradiation, particularly in nonsem inomas. 57 T he addition of chem otherapy has been asso ciated with som e of the reported sarcom as, and there is a risk of acute leukaem ia and bladder cancer following som e cytotoxic regim es. 63 C ombined treatment with radiotherapy and chem otherapy is rarely practised, unlike the situation with H odgkin' s disease where cure rates m ay be improved with optim um use of both m odalities. Lowdose radiotherapy (16± 20 Gy) is recomm ended for in situ carcinom a of the contralateral testis after positive biopsies, and can prevent the developm ent of invasive germ cell tum ours, 64 dem onstrating the general acceptance that the bene® ts of irradiation far outweigh any risk of radiation-induced cancer.
Risk factors other than radiation have been identi® ed in some case reports of radiation-induced sarcom as in testicular cancer patients, and it has been postulated that som e teratom as transfo rm into sarcom as spontaneously. 65,66 Both radiation and chem otherapy can produce differentiation of im m ature or undifferentiated teratom as.

Brain tum ours
Radiotherapy has been used in the treatm ent of benign brain tumours in large num bers of patients  69 but these too are rare and may be dim inishing in fre-quency w ith m odern radiotherapy techniques and equipm ent.

Head and neck cancers
Excluding retinoblastom as, the risk of sarcom as following therapeutic irradiation in head and neck cancers is very low. In the 10 reports totalling 14 000 patients, sum m arized in Table 5, 12,72± 80 the incidence was 0.16%. This is consistent with the one case of radiation-induced sarcom a for every 1250 treated patients estimated by Parsons in a

C arcinoma of cervix
Endocavitary and external beam radiotherapy have long been established as a standard m ethod of treating cervical cancer. In 1985, an international collaboration involving 15 cancer registries in Europe and N orth Am erica reported on the num bers of second cancers am ong a huge cohort of 182 000 wom en treated for cervical cancer, com paring those treated by radiotherapy and by surgery alone. 82 Connective tissue tumours were slightly increased in the irradiated group (relative risk 2.3 after 10 years), as were som e other m alignancies, including carcinom as of the bladder, rectum and other genital organs in heavily irradiated tissues, but not bone tum ours. Sarcom a sites were not identi® ed and the num ber occurring away from the radiation port w as not stated. O verall, the num ber of cancers which could be attributed to irradiation was estimated to be at m ost 5%, and this was m ore than counterbalanced by the reduction in the incidence of breast cancer, probably secondary to ovarian ablation. T hree of four sm aller studies showed sim ilar trends. 83± 86 T able 6 show s the reported cases of sarcom as in 170 000 long-term survivors of radiotherapy, for an overall incidence of 0.08%.
In a follow -up study from the International Radiation Study of C ervical C ancer, Boice et al. analyzed 38 sarcom as according to site using case-m atched controls and found no signi® cant increase in the risk for bone sarcom as but som e increased risk for soft tissue sarcom as (both relative risks 1.9) w hich occurred within the irradiated region. 87 T he risk of pelvic soft tissue sarcom as was doubled following cervical radiotherapy. D espite m any anecdotal reports of uterine tum ours including sarcom as occurring after radiotherapy, 88,89 a strong body of epidem iological literature has not identi® ed a de® nite relationship betw een radiotherapy and sarcom as arising from the uterus. 90

C onclusions
Radiotherapy has an established place in the treatm ent of a wide range of neoplastic conditions, and the proportion of cured patients continues to grow. Virtually all will enjoy a further 10 years of life before they approach the time at which any sarcoma induced by their earlier therapy could appear. D espite num erous anecdotal reports of sarcom as arising in cancer patients who have received radiotherapy, large epidemiological studies involving hundreds of thousands of radiotherapy patients have shown that at m ost only 5% of all second primary cancers can be convincingly linked to the radiation treatment, 91 and all the others are attributable to life-style, inheritance and other carcinogens. O f the 360 000 radiotherapy patients evaluated in the reports in this review, only 0.1% or one patient in a thousand has subsequently developed sarcom as which m eet the criteria of being radiation induced.
The risk of sarcom a developm ent following irradiation is extremely low and should be w eighed against the risks of death and carcinogenicity from horm onal anti-cancer agents (e.g. cardiovascular toxicity and vaginal adenocarcinom a from stilboestrol, breast cancer from oestrogens, endom etrial cancer from tam oxifen) or chemotherapeutic drugs (e.g. leukaem ia from alkylating agents and etoposide, bladder cancer and possib ly sarcom a from cyclophosph am ide 92 ), or the operative m orbidity and m ortality of surgery, all alternatives to radiotherapy in the treatm ent of cancer patients. 15 Potential carcinogenicity should not be regarded as a contraindication to the use of radiotherapy, even in patients with retinoblastoma and neuro-® bromatosis w here a higher risk of tum our induction is recognized. In the m ore com m on cancers as w ell, the long-term bene® t derived from radiotherapy far outweighs the serious side-effe cts, provided standard techniques are used.
Radiation-induced sarcom as often arise in areas heavily dam aged by radiation doses far in excess of norm al tissue tolerance and they are likely to be seen even less frequently with current standards of treatm ent which use lower doses of m egavoltage ir-radiation delivered in daily fractions of 2 Gy, avoid m ultiple courses of treatment and utilize planning by m odern com puter-assisted techniques, resulting in acceptable acute toxicity and infrequent chronic bone and soft tissue com plications. Another m uchfeared potential com plication, the anaplastic transform ation of benign or low-grade tum ours follow ing irradiation, occurs even less frequently than radiation-induced sarcom as, and cases identical to those anecdotally attributed to irradiation have also been found follow ing other types of treatment in the absence of radiotherapy. 93± 95 There is evidence in recent reports of radiationinduced sarcom as that aggressive management with early diagnosis and com plete local resection results in prolonged survival no different from sarcom as unrelated to prior irradiation, particularly if they develop in soft tissue rather than bone. 23,30,36± 38,96 Regular follow-up exam inations of cancer patients at the centre where their treatment took place should prom ote early detection by prom pt investigation of any new m asses, with biopsies to exclude the alternative diagnosis of relapse from the original cancer which com m only delays optim al sarcom a treatment. C om plete excisions should be perform ed and consideration given to post-operative re-irradiation as occurs for prim ary soft tissue sarcom as. A role for chem otherapy in treatment is yet to be established, and som e agents may increase the incidence of these rare tum ours. 21,91