The Science of Soy: What Do We Really Know?

A stroll through nearly any American grocery store or pharmacy yields ample proof of the soybean’s increasing role in the U.S. diet. Food packaging offers statements about products’ soy content and the purported associated health benefits. Products such as tofu, soy milk, soy-based infant formula, and meatless “texturized vegetable protein” burgers are widely available. Shelves of dietary supplements and nutraceuticals are stocked with isoflavones, naturally occurring estrogenic compounds found in soy. The general impression is one of certainty that both soy and soy isoflavones deliver many health benefits, including prevention of cardiovascular disease, cancer, and osteoporosis, as well as treatment of menopausal symptoms. The science is less absolute, however, and still evolving. 
 
Soy provides a complete source of dietary protein, meaning that, unlike most plant proteins, it contains all the essential amino acids. According to the American Soybean Association, 3.14 billion bushels (85.5 million metric tons) of soybeans were harvested in the United States in 2004. Approximately half of the harvest was exported, and most of the remainder was crushed to produce oil and protein meal for domestic use. An April 2006 report from the USDA Economic Research Service indicates that only a small amount of whole soybeans are used to produce soy foods, and just 2% of soy protein meal is used for human consumption; the rest is used for animal feed. 
 
The Soyfoods Association of North America reports that U.S. sales of soy foods reached $3.9 billion in 2003, continuing an 11-year trend of 15% average annual increases. According to the United Soybean Board’s 2004–2005 Consumer Attitudes About Nutrition report, 25% of Americans consume soy foods or beverages at least once per week, and 74% view soy products as healthy. 
 
Nevertheless, Americans as a whole still consume very little soy protein. Based on 2003 data from the UN Food and Agriculture Organization, per-capita soy protein consumption is less than 1 gram (g) per day in most European and North American countries, although certain subpopulations such as vegetarians, Asian immigrants, and infants fed soy-based formula consume more. The Japanese, on the other hand, consume an average 8.7 g of soy protein per day; Koreans, 6.2–9.6 g; Indonesians, 7.4 g; and the Chinese, 3.4 g. 
 
Traditional soy foods include tofu, which is produced by pureeing cooked soybeans and precipitating the solids, and miso and tempeh, which are made by fermenting soybeans with grains. “Second generation” soy products involve chemical extractions and other processing, and include soy protein isolate and soy flour. These products become primary ingredients in items such as meatless burgers, dietary protein supplements, and infant formula, and are also used as nonnutritive additives to improve the characteristics of processed foods.

Summary The incidence of a new primary non-germ cell malignancy was determined in 876 patients with testicular cancer treated at the Norwegian Radium Hospital from 1956 to 1977. Sixty-five patients developed a second cancer leading to a statistically significant increased relative risk (RR = 1.58), especially if extended radiotherapy had been given (RR = 4.13). The excess risks of developing lung cancer and malignant melanoma were 2.03 and 3.89, respectively. Increased RR for these two cancer types were seen both after extended radiotherapy and after radiotherapy combined with chemotherapy. Studies of the time between treatment and secondary lung cancer indicated that the development of the new lung cancer could be partly treatment related, whereas the raised incidence of malignant melanoma may be related to the frequent health checks performed in patients with testicular cancer. Patients who had received extended radiotherapy were also at an increased risk of developing cancer of the stomach and of the colon. Three cases of acute leukaemia were observed more than 5 years after treatment, all of them in patients who had received abdominal radiotherapy only. It is concluded that patients apparently cured of a testicular cancer have an increased risk of developing a new treatment related non-germ cell malignancy, in particular lung cancer. The application of the extended radiotherapy or the combination of radiotherapy and chemotherapy containing alkylating drugs should be avoided in order to reduce this excess risk. Testicular cancer represents 1.25% of all male malignancies in Norway. Ninety per cent of the patients with testicular cancer are now cured (Peckham, 1988;Fossa et al., 1988). As the mean age of patients with this malignancy is about 35 years, most of them will live for 30-40 years after treatment. The problem of treatment related long-term toxicity has therefore become important. Some authors have described organ-related side-effects occurring 5-10 years after radiotherapy and/or chemotherapy, such as nephrotoxicity, neurotoxicity, pulmonary, cardiovascular and gonadal toxicity (Roth et al., 1988;Fossa et al., 1988;Hansen et al., 1988;Aass et al., 1990). Four larger series have dealt with the incidence of second non-germ cell tumours observed after treatment of testicular cancer (Cockburn et al., 1983;Hay et al., 1984;Kleinermann et al., 1985;Kaldor et al., 1987). In these large studies the relation to treatment is, however, not always considered specifically. The objective of the present paper is to study the incidence of new non-germ cell malignancies in testicular cancer patients who received radiotherapy.

Material and methods
Cancer registration and statistical analysis During the period 1956-1977, 1,484 cases of cancer of the testis were diagnosed in Norway. Of these patients, 68% had their primary treatment at the Norwegian Radium Hospital (NRH). The present study concerns this group of patients. One hundred and twenty-six patients were excluded because they had not received standardised radiotherapy. Among these, chemotherapy had been given to 64 patients, while most of the remaining 62 patients had no further treatment after orchidectomy. Two other patients were excluded because of a malignant tumour before the diagnosis of testicular cancer. The final series thus consisted of 876 patients.
All new cases of cancer in Norway have been recorded by the Cancer Registry since 1953. This is based on compulsory reporting by hospital departments and histopathological laboratories. All death certificates are coded by the Central Bureau of Statistics and information about all deaths is passed on regularly to the Cancer Registry.
From the census in 1960 a personal identification number has been allocated to all inhabitants of Norway. This number was used for matching all second primaries. The matching was automated after 1960 and manual for the preceding years.
A standard life-table procedure was used to calculate person years at risk and expected number of cancer cases. For estimation of the expected number of cancer cases the 5-year age-specific incidence for each of the years 1957-1987 for the whole country was used. The study was based on a comparison of observed and expected incidence of cancer for the period 1957-1987 (relative risk (RR): observed/expected cancer cases). Ninety-five per cent confidence intervals were determined by assuming a Poisson distribution of the observed number of cancer cases. A result was regarded as statistically significant if the 95% confidence interval did not include 1.00.
The follow-up of the patients started 1 year after the date of diagnosis and all patients were followed up until the end of 1987 or to the middle of the year of death or emigration. During the first year after the diagnosis of testicular cancer none of the 876 patients died, and one cancer case was observed against 0.1 expected.
The medical records of the 876 patients were reviewed for diagnostic and treatment details. The differentiation between seminoma and non-seminoma was based on the routine histological evaluation done by members of the Department of Pathology at the NRH (Table I). Retrospective staging was based on the Royal Marsden classification system (Peckham et al., 1979).

Treatment
Radiotherapy A detailed description of the treatment principles has been given elsewhere (Fossa et al., 1988). High-voltage radiotherapy represented the main treatment modality. Radiotherapy was given by Betatrons 31 or 33 MV (1956)(1957)(1958)(1959)(1960)(1961)(1962)(1963)(1964)(1965)(1966)(1967)(1968)(1969) or by linear accelerators (5-8 MV) (1970)(1971)(1972)(1973)(1974)(1975)(1976)(1977). Patients with stage I disease received abdominal radiotherapy. The fields included the bilateral para-aortic lymph nodes and the ipsilateral iliac lymph nodes ( Figure 1). The daily dose was 2 Gy. Five fractions were given weekly. If radiotherapy was delivered by Betatrons the iliac lymph nodes were irradiated by an anterior field only (field size approximately 12 x 15 cm), whereas the para-aortic lymph nodes were covered by a posterior field (field size approximately 12 x 15 cm). Both fields were treated daily. If linear accelerators were used, radiotherapy was given to an anterior and posterior abdominal field (L-field) and one field was treated daily. Seminoma patients were routinely treated with a total dose of 36-40 Gy; non-seminoma patients received 50 Gy to the abdominal fields. In patients with stage II and stage III tumours, additional radiotherapy (30-40 Gy) was given to mediastinal fields, including the left or both supraclavicular fossae (Figure 1).
For the purpose of this analysis the mid-plane dose to the mediastinum from scattered irradiation was estimated. In patients receiving 40 Gy from a Betatron to standard abdominal fields, the mediastinal dose was <20 cGy. The comparable dose in patients treated by a linear accelerator was 60-70 cGy. In the latter cases the superficial layers of the skin of the thorax were exposed to a dose of 100-120 cGy.
Chemotherapy Chemotherapy was given mainly to patients with stage IV disease or as secondary treatment in case of relapse.
The type of the cytostatic drugs and of the chemotherapy regimens has varied during the years. During the first 10 years cyclophosphamide was given as a single drug. During the years 1966-1975 the Li regimen (Li et al., 1960) and mithramycin (Klepp et al., 1975) were added to the therapeutic armamentarium. From 1975-1977 adriamycin based combination chemotherapy was the treatment of choice whenever systemic chemotherapy was considered (Klepp et al., 1977).

Results
For the total series there was a significantly raised relative risk for the development of a new non-germ cell cancer (RR: 1.58), with a statistically significant excess risk for lung cancer and for malignant melanoma (Table II). In addition, cancer of the stomach, colon and bladder tended to occur more frequently than expected. Three cases of leukaemia were observed against 0.85 expected in the subgroup of patients with radiotherapy only. If the different subgroups were considered, lung cancer and malignant melanoma were again the most frequent new malignancies, but only in patients who had been treated with extended irradiation or combined radio/chemotherapy. In the evaluable patients, half of the malignant melanomas were inside the irradiation field, the others were not. All malignant melanomas were on the trunk. The RR for lung cancer was also increased (although not significantly) in patients who only received abdominal irradiation. There was a higher incidence of bladder cancer and cancer of the stomach in patients after abdominal radiotherapy, but this was not statistically significant.
In general, the RR of a solid malignancy was highest 5-14 years after treatment for testicular cancer (Table IV). The RR for malignant melanoma was highest within years 1-4 and decreased thereafter. Two of the three cases with leukaemia developed 15 years or later after the diagnosis of testicular cancer. Due to the low number of cases no definite statement about time dependency can be made for the other malignancies.

Discussion
In the present study we have not analysed the incidence of subsequent contralateral testicular germ cell tumours, as these data are incompletely recorded in the Cancer Registry. However, other studies have shown that there is an excess risk of a new primary testicular cancer in patients surviving their first germ cell cancer (Dieckmann et al., 1986;von der Maase et al., 1986). These second germ cell tumours probably develop on the basis of in situ lesions in the remaining testicle (von der Maase et al., 1986). This report confirms the increased incidence of new primary non-germ cell malignancies in patients with testicular cancer, as also found by Kaldor et al. (1987), Hay et al. (1984, Kleinermann et al. (1985) and Cockburn et al. (1983). The incidence of a new cancer was highest within the groups of patients who received extended radiotherapy or combined radio/chemotherapy. Such intensive treatment represents a particular risk factor for second cancer development also seen in patients cured from Hodgkin's disease (Toland et al., 1978). However, patients who received radiotherapy alone also showed an increased risk of second solid malignancies, especially lung cancer, malignant melanoma and cancer of the stomach.
We found an excess risk of lung cancer like Hay et al. (1984) but unlike Kaldor et al. (1987). This increased risk was most evident 5-14 years after the diagnosis of testicular cancer. Combined radiotherapy/chemotherapy or extended irradiation seemed to yield a particularly high risk. The incidence of lung cancer has been linked epidemiologically to gamma radiation exposure (Smith & Doll, 1982;Kato & Schull, 1982). As in these studies, the majority of the lung  (Kaldor et al., 1987). A genetic predisposition may account for the increased risk of a second malignancy in testicular cancer patients. It is generally believed that cancer develops by multiple steps resulting in changes in the growth control mechanisms. The description of genetic events in the development of retinoblastoma (Cavanee et al., 1983), based on the Knudson two-hit model (Knudson, 1971), initiated an intense search for tumour suppressor genes. Loss of specific DNA sequences in tumour cells have been shown for several familial and sporadic solid malignancies (Ponder, 1988). Site-specific allele losses on chromosome 3p have been shown for renal cell carcinoma and for lung cancer (all types). Sequences on chromosome lip are lost in Wilms' tumour and bladder cancer as well as in certain types of lung cancer (Willey et al., 1988). Recently we have shown that both these regions also are involved in testicular germ cell tumours (Lothe et al., 1989).
The fact that both the 3p and lip chromosomal regions are shown to be involved in lung cancer strengthens the possibility of a genetic predisposition in a subset of testicular cancer patients who develop lung cancer after treatment. A genetic predisposition may be suggested by features  like familiar occurrence of testicular neoplasms (Gedde-Dahl et al., 1985), bilateral tumours and multiple primary malignancies. The excess risk of malignant melanoma is intriguing. Radiation exposure within the treatment fields or by scattered irradiation to the trunk elsewhere may represent one explanation. However, the highest incidence of malignant melanoma occurred as early as 1-4 years after the diagnosis of testicular cancer, making a relationship to treatment less probable. The increased incidence of malignant melanoma is partly due to the increased medical attention during the frequent follow-up examinations which testicular cancer patients undergo. As for lung cancer a common predisposing factor might be present. Hay et al. (1984) described an excess risk of skin cancer after the diagnosis of testicular cancer, not distinguishing between non-melanoma and melanoma. However, from these authors' discussion it becomes evident that most of the second skin cancers were non-melanoma registered in patients who (due to their primary testicular cancer) had more frequent and intensive health examinations than the general population.
An increased risk of transitional cell carcinoma in irradiated sites has been reported previously (Hay et al., 1984;Kleinermann et al., 1985). This observation is partially confirmed in the present study by an increased RR of second bladder cancer, but again the hypothesis of the common predisposition cannot be rejected.
For the other solid tumour types the numbers are small and do not allow interpretation. However, in the future the incidence of a new cancer of the colon and stomach cancer should be evaluated in larger series. Kaldor et al. (1987) demonstrated an excess risk of rectal cancer after treatment for testicular cancer.
Only three cases of leukaemia were observed; all were acute leukaemia and all three patients had received abdominal radiotherapy as their only treatment. The RR was not significantly increased as compared to the overall incidence of leukaemia in the general population. However, for acute leukaemia there was a significantly increased RR among our patients. This is in line with observations of Kleinermann et al. (1985) andRedmann et al. (1984), who found a statistically significant excess risk of acute leukaemia in patients treated for testicular cancer, after irradiation alone, after chemotherapy alone or after a combination of both treatment modalities. On the other hand, Hay et al. (1985) did not find an excess risk of acute leukaemia in irradiated testicular cancer patients. In the literature it is generally quoted that radiation or chemotherapy induced leukaemia usually occurs within 2-3 years after completion of treatment. However, in our three patients with acute leukaemia the malignancy was diagnosed 5 years or more after the treatment for testicular cancer (more than 15 years after in two patients). This observation makes any treatment relation less probable.
The introduction of cisplatin into the treatment of testicular cancer has dramatically changed the treatment policies in testicular cancer. Non-seminoma patients without metastases no longer receive adjuvant radiotherapy. In testicular cancer patients with metastases adriamycin or cyclophosphamide are used rarely. Cisplatin-based chemotherapy represents the principal therapy. Whether cisplatin-based chemotherapy increases the risk of secondary cancer is unknown. Alkylating agents, such as iphosphamide, are, however, still applied extensively in the modern therapy of both non-seminoma and seminoma, and abdominal radiotherapy is still the treatment of choice in low stage seminoma. The combination of chemotherapy with radiotherapy represents an actual therapeutic alternative in advanced seminoma. All these treatment modalities may increase the risk of a second non-germ cell malignancy in surviving patients.
We feel that the present series allows the following conclusions which are still relevant today. 1. The excess risk of a new non-germ cell cancer in the group of patients with extended radiotherapy or combined radiotherapy/chemotherapy should lead to reluctance to apply these treatment modalities routinely in testicular cancer patients, in particular if alkylating agents and/or adriamycin are applied. Such combination treatment should only be given if strongly indicated. 2. Due to a probable excess risk of lung cancer, the young testicular cancer patient should be warned against avoidable exposure to known carcinogens. In particular, he should be strongly advised not to smoke. 3. As our figures for an increased RR for some new cancers (bladder, stomach, colon) are only suggestive, and do not yield statistically significant differences, large co-operative studies are needed to confirm or disprove the observation. Such studies should take into account the different treatment modalities.