Cancer patient Diane Rees keeps a photocopy of her latest ultrasound report in a plastic wallet and flourishes it like a certificate of airworthiness. That report of 2 September says her liver is normal: no sign of the tumour diagnosed in March last year.

It is the second time Ms Rees, 50, has confounded both the experts and the statistics. In October 1991 she was found to have multiple tumours in her brain, one in the vital brain stem area. All have since vanished, leaving just one small nodule.

In brain and liver, the tumours were metastatic or secondary. Once a cancer has spread beyond its original site, the outlook is extremely poor in most cases.

Christopher McKenzie, consultant in clinical oncology at Hammersmith Hospital, London, who treated Ms Rees as a private patient at the Cromwell Hospital, cannot think of two more potentially fatal sites for cancer, and describes her recovery as 'doubly miraculous'.

Her specialists have still not located the original cancer from which her brain and liver tumours developed. This puts Ms Rees in the 2 to 12 per cent of cancer patients who are designated MCPSU - metastatic cancer, primary site unknown. In about 16 per cent of patients with MCPSU, the primary source is never found, even at autopsy; usually it is too small to be located by manual or imaging examinations.

Of those patients with a similar initial diagnosis, about half will survive for five to six months, and 3 to 7 per cent will survive for five years. Dr McKenzie says Ms Rees's condition put her so far at the lower end of these figures that her chances of surviving for any appreciable period were nil. When he first saw her two years ago, he gave her only weeks to live.

His reaction to last month's liver scan was one of 'surprise and delight'. Ms Rees says: 'He grabbed hold of me and said he'd never seen such a thing before.' Her next chemotherapy treatment was cancelled, and two weeks ago her daily steroid intake, which helps to guard against any reactivation of the brain tumours, was halved.

Dr McKenzie cites two possible reasons for this outcome. First, he increased the dose of radiation to the brain tumours after Ms Rees responded well to the radiotherapy. 'Other than that, she had no special treatment.' But, more important, he said, was the attitude of his patient. 'She wants to live, and has always accepted that the treatment has to be done. All the staff love her - she is so easy to deal with.'

The one person who was not surprised by the disappearance of the tumours was Ms Rees herself. 'I was certain that it was never going to kill me. I'm someone who usually sees the bright side of life.' She sought none of the cancer counselling services on offer, nor any alternative treatments.

Until the onset of her illness, Ms Rees was a legal secretary. Divorced many years ago, she lives in West Hampstead, north-west London, where she and her three cats share the maisonette of a platonic lawyer friend, John.

She was a private patient, not because of vast wealth but because she maintained payments on a policy bought for her by a previous employer. 'Last year my premium was pounds 486. It's been a great investment.'

Ms Rees noticed unusual breathlessness in October 1991. Then her left leg began to drag occasionally, and her left arm became weaker. For three weeks she visited her GP almost daily. Rheumatism was diagnosed, and she was prescribed a drug that made her feel sick.

She felt that her doctor was beginning to find her a nuisance, and when he learnt of her private health policy, he sent her to a private consultant. By this time, she had lost control of the left side of her face.

Dr Ann Coxon, a Harley Street neurologist, immediately arranged for a magnetic resonance imaging scan. 'As I arrived next morning for my scan,' Ms Rees says, 'the doctor told me she was relieved to see me - she had given me only 24 hours. The scan showed what looked to me like two great bullet holes through my head. (These were her brain tumours, one in the nerve centre of the brain, which accounted for the loss of use of the left side of her body.) Dr Coxon told me it was inoperable and would be terminal. She was very straight with me. I rushed to the loo to adjust my contact lens and broke down. But I didn't cry for ever.'

'Diane doesn't give a damn about statistics,' Dr Coxon says. 'She said from the beginning she knew she'd beat it. The day after I told her, she was up and running.'

The day before radiotherapy began, Ms Rees was told to shave her head. She bought a wig and went to her normal stylist at Vidal Sassoon, who cut her hair to one-eighth of an inch; she was not charged. Later, when her 'beautiful hair' was mistakenly admired by a receptionist, Ms Rees pulled the wig off in front of her. 'Losing my hair has never worried me,' she says.

In November 1991, she began a three-month session of radiotherapy, followed by a period of increased radiation. During the first phase, a third brain tumour appeared near her upper neck. This allowed for a biopsy, which confirmed the tumours' malignancy.

But at no point did Ms Rees despair. 'I got such backing from Dr Coxon and Dr McKenzie that I had total confidence in them,' she says. She did, however, make a will - and she has continued to smoke.

That Christmas, she had 13 people for lunch. As the steroids she was taking made it hard for her to sleep, she stayed up all night cooking. 'It's the first time I've had Christmas lunch ready on time.'

In March 1992, Ms Rees reported pains in her right leg, 'as if the ligaments were stretching'. Dr Coxon gave her exercises and asked to see her again 10 days later. The pain moved to her waist and became continuous. On re-examining her, the doctor immediately diagnosed liver cancer. Two days later, chemotherapy began.

Statistically, Ms Rees's chances of survival now dropped dramatically, her doctors say. Though she coped well with the chemotherapy, it was not plain sailing. In May, after 14 months of cancer drugs, staff were having problems inserting needles into her veins, so a Hickman line was inserted into the main artery of her heart. Three months later, she had an allergic reaction to the chemotherapy. 'My hands itched so badly, I was jumping up and down. Dr McKenzie told the nurse to disconnect it immediately.'

Then, when her liver was scanned again, came the remarkable news: the tumour had vanished in the past month.

Dr McKenzie says: 'Of all the people who get metastases in the brain from an unknown primary, few survive. It was one of the worst cases I've ever seen. Once a cancer has spread beyond the tissue in which it was originally located, it is usually fatal. I was surprised when I saw her again with the liver tumour, and even more surprised when the scan showed it had disappeared.

'It is an unusual case. She was young, and the sites in the brain appeared suddenly. Patients whose disease goes contrary to expectations, sometimes respond unexpectedly to treatment.' But generalisations should not be made from Ms Rees's case, he says, and the disappearance of the metastases does not mean she is fully recovered.

Dr Coxon is more sanguine. 'We don't treat patients without the knowledge that some benefit to an extraordinary degree. I told her that she must hope that she is one of them, although I also told her to live her life day by day. She had a distinct belief in herself, which must have been a factor. To identify winners is very difficult, but one notices that they have a deep sense of their own powers of recovery.'

Ms Rees has been living on state benefit, and has just begun to receive an allowance for the over-50s who are terminally ill. 'It's the old- age pension they are paying early,' she says. 'But they're going to be paying it for a long time to come.'

(Photograph omitted)