Prior research has suggested that vitamin D levels may be associated with breast cancer risk, an idea that is biologically plausible.

"We know from basic science studies that breast cancer cells have vitamin D receptors and can interact with vitamin D," explained Goodwin, who plans to present her findings at the ASCO annual meeting, being held from May 31-June 3, in Chicago.

There is, however, a lot of confusion about the different health benefits of vitamin D and, the authors said, daily allowance recommendations vary greatly around the world.

The study involved 512 women newly diagnosed with localized breast cancer (confined to the breast and arm pit) between 1989 and 1995. All participants had had blood taken at the time of diagnosis and had also filled out a questionnaire on diet. Vitamin D levels were measured by radioimmunoassay.

The women, whose age averaged about 50, were followed for just under 12 years. Of the total, 37.5 percent were deficient in vitamin D (the lowest levels), 38.5 percent had insufficient levels of vitamin D (not deficient but not quite healthy levels), and 24 percent had levels in the healthy range.

Women who were premenopausal, weighed more, had high insulin levels and had more aggressive tumors were all more likely to have low vitamin D levels.

"Fat tissue acts as a trap for vitamin D," Goodwin explained. "Levels were also lower in younger women, which was a bit of a surprise, until we realized older women were taking more supplements."

The risk of breast cancer spreading was almost double in women with deficient levels of vitamin D at diagnosis, compared with women whose levels were healthy, the researchers said.

The risk of dying of breast cancer was 75 percent higher in women with too-low levels of vitamin D versus women in the optimal range, they added.

However, there was no survival difference between women with healthy levels and insufficient levels. The data also suggested that there may be a small increased risk of metastasis or death if vitamin D levels are high, but not a statistically significant effect, Goodwin said.

"This suggests that there's a healthy level for vitamin D and, if you are deficient, you have an increased risk of metastasis, but if you go above [a certain point], your risk of death goes up again," Goodwin said.

Although it's not a firm conclusion, Goodwin added, the finding "is raising a red flag. It's probably not safe to take very high levels and get vitamin D levels up really high."

And the findings are consistent with another study which found that men with prostate cancer who took vitamin D died earlier than those taking a placebo.