A number of reports of variations in post-mortem pineal weight as a function of cause of death have been summarized by Tapp (88). Of the most interesting, hypoplasia of the pineal in association with retinal disease may be causally interrelated. Tapp has reported that pineals in patients dying of carcinoma of the breast or melanoma are heavier than those from patients with other cancers. Very large pineals (1 g) have been described in a rare genetic syndrome with insulin resistance (89). Sudden infant death syndrome (SIDS) is associated with small pineals and decreased melatonin production (90). SIDS deaths usually occur at night and may be associated with abnormalities of sleep. If melatonin helps to coordinate circadian organization in the developing infant, its underproduction may contribute to the disorder.

Tumors of the pineal region in children are frequently associated with abnormal pubertal development (91). In precocious puberty it was thought that the capacity of the pineal to inhibit sexual development was impaired. Much evidence now suggests that precocity is due to the production of human chorionic gonadotrophin (beta-hCG) by germ cell tumors of the pineal (92, 93). Delayed puberty has also been associated with pineal tumors. Pineal tumors are heterogeneous and may arise from germ cells (teratomas, germinomas, choriocarcinomas, endodermal sinus tumors, mixed germ cell tumors), pineal parenchymal cells (pineoblastoma and pineocytoma), and the supporting stroma (gliomas) (94). All are rare (less than 1 per cent of intracranial space-occupying lesions) and tend to occur below 20 years of age with the exception of parenchymal cell tumors, which occur equally in adults and children. Germinomas respond well to radiation therapy, whereas primary surgery is more frequently the treatment of choice in other types. Tumor markers in CSF, alpha-fetoprotein and beta-hCG, together with CSF cytology and imaging (CT or MRI), aid in differential diagnosis. The most common symptoms are secondary to hydrocephalus (headache, vomiting, and drowsiness) together with the triad of visual problems, diabetes insipidus, and reproductive abnormalities (94). Germinomas and teratomas occur predominantly in males. Precocious puberty is more commonly associated with teratoma. As beta-hCG is identical to beta-LH, pubertal development can be directly attributed to ectopic beta-hCG production in many cases. Moreover, the predominance in boys may be explained on the basis that LH alone can stimulate testosterone production, whereas in girls both LH and FSH are required for ovarian follicular development and estrogen production. Reviewing a series of 37 patients, Drummond and Rosenfeld (95) concluded that there have been significant improvements in outcome over the last 30 years. A 5 year survival of 62% was quoted for germinomas, but only 14% for other malignant tumours.

Classification of pineal parenchymal tumours is complicated by the presence of mixed pineocytoma-pineoblastoma types some with intermediate differentiation. A new classification has been proposed based on histological features which is closely related to patient survival (96).

There is no consistent information on overproduction or underproduction of melatonin with specific types of tumor. Recent work suggests that melatonin is absent or very low in treated or untreated pineal germinomas, but the consequences remain to be defined (97).

In tumour-bearing animals both increases and decreases in melatonin production have been reported. In humans low levels may be associated at least with (stage-dependent) breast and prostatic cancer (98) (and some other endocrine-independent tumours) with a negative correlation to tumour size. Remission is associated with normalisation of melatonin levels. In ovarian cancer, on the other hand, elevated melatonin is reported. A number of broad studies that have included various oncological conditions report significant differences, both increases and decreases, in plasma melatonin between types of cancer and control populations. At present these are uninterpretable, and no mechanisms has been shown to account for the observed changes. The subject has been extensively reviewed recently (98). Some data suggests that the growth of human benign prostate epithelial cells depends on both steroids and melatonin (99).

Considerable effort has been expended investigating melatonin timing and production in prospective and retrospective ‘field’ studies of cancer patients and shiftworkers (women shiftworkers have increased risk of breast cancer) assessed by the urine levels of aMT6s. An increased risk of breast cancer has been attributed to lower melatonin; however, the data are inconsistent and in some cases may be interpreted as a different timing of the rhythm rather than altered production (100-102). Most data are based on aMT6s concentration in morning void urines, where a change in timing of the rhythm can lead to under or over estimation of production. Breast cancer risk is associated with the presence of a clock gene polymorphism (in hper3) (103) which is associated with morning diurnal preference (lark) (104), which in turn is associated with earlier timing of the melatonin rhythm and hence lower morning aMT6s values. When urine collected over 24h is used for aMT6s measurement no associations have been found in 2 prospective studies (102, 105). Thus, a causal connection has not been established and many other factors (such as continual disruption of the circadian system in general) may be involved.

The association of both breast cancer and childhood leukemia with environmental exposure to electromagnetic fields (EMF) has also been attributed to melatonin suppression by EMF (68). There is little convincing evidence for this association and most recent data deny any acute suppression of melatonin in humans by EMF.

Melatonin has been extensively used in psychiatry to assess biological clock status. There is evidence for a decline in amplitude of the melatonin rhythm in depression associated with an increase in cortisol, and possibly an increase in mania, although not all studies are consistent (see for references (106). Exceptionally delayed melatonin rhythms in winter in patients with SAD compared with the small delay seen in normals (107, 108) have been reported, and Parry and co-workers have found abnormal melatonin patterns and response to light in premenstrual dysphoric disorder. At present there is still no consensus as to what causes SAD. Light treatment for SAD appears to be slightly more efficient when given in the morning (albeit with a large placebo effect) (109), thereby inducing an advance in the melatonin rhythm. However, other mechanisnms are also possible. Many pharmacological antidepressant treatments stimulate melatonin secretion, acting through increased availability of the precursors tryptophan and serotonin and the major pineal neurotransmitter norepinephrine, or by direct action on serotonin and catecholamine receptors . There may be a link between an increase in melatonin production and efficacy of treatment, and this possibility merits exploration. A recently introduced melatonin agonist is at present being developed for anti-depressant activity (110).

Many clinical attempts have been made to relate circulating melatonin to endocrine and other pathology. The results on the whole are difficult to interpret and inconsistent (see references (12, 21)). Liver disease such as cirrhosis, which impairs metabolic function, leads to higher than normal plasma concentrations of melatonin (111). Drugs that stimulate or suppress hydroxylation and conjugation mechanisms or that compete for metabolic pathways can be expected to affect circulating melatonin. Surprisingly, little evidence exists for a disturbance of melatonin secretion in major sleep disorders such as narcolepsy and Klein-Levine syndrome and in delayed sleep phase insomnia delays in the rhythm are not always found.