rss
Emerg Med J 21:e3 doi:10.1136/emj.2003.008870
  • Electronic pages

Value of serum prolactin in the management of syncope

  1. S Ahmad1,
  2. M W Beckett2
  1. 1Paediatric Accident and Emergency Department, Queen Elizabeth Central Hospital, Blantyre, Malawi
  2. 2Accident and Emergency Department, West Middlesex University Hospital, Isleworth, Twickenham, UK
  1. Correspondence to:
 Dr M Beckett
 Accident and Emergency Department, West Middlesex University Hospital, Isleworth, Twickenham TW7 6AF, UK; djshaweaol.com
  • Accepted 25 June 2003

Abstract

Objective: A meta-analysis of studies of the usefulness of raised serum prolactin in diagnosing generalised tonic-clonic seizures (GTCS) in patients presenting to the accident and emergency (A&E) department after a single episode of syncope.

Methods: A three part question was defined. Medline, EMBASE, PubMed, the Cochrane Library were searched to identify relevant studies. Studies were evaluated for eligibility and quality and data extracted to calculate sensitivity (SN), specificity (SP), and likelihood ratios (LR).

Results: Of 13 relevant studies only three met the criteria for evaluation. If a serum prolactin concentration is greater than three times the baseline when taken within one hour of syncope, then in the absence of test “modifiers”: (1) the patient is nine times more likely to have suffered a GTCS as compared with a pseudoseizure positive LR = 8.92 (95% CI (1.31 to 60.91)), SN = 0.62 (95% CI (0.40 to 0.83)), SP = 0.89 (95% CI (0.60 to 0.98)) and (2) five times more likely to have suffered a GTCS as compared with non-convulsive syncope positive LR 4.60 (95% CI (1.25 to 16.90)), SN = 0.71 (95% CI (0.49 to 0.87)), SP = 0.85 (95% CI (0.55 to 0.98)).

Conclusion: A positive test result is highly predictive of a GTCS, however a negative test result does not necessarily exclude a seizure. Serum prolactin should be measured in patients presenting to the A&E department within an hour of a syncopal episode, unless the cause is immediately obvious.

A 24 year old patient suffers a single episode of syncope. Clinical evaluation reveals no obvious cause or important sequelae. Is a serum prolactin concentration useful in excluding a seizure disorder?

BACKGROUND

Transient loss of consciousness is a common and diagnostically challenging problem for the emergency physician. The clinical picture may be suggestive of a genuine seizure disorder or syncope. Syncope as a presenting complaint accounts for up to 3% of emergency department visits and up to 6% of acute medical admissions.1 Even after extensive investigation, up to 41% of patients may not have a cause identified.2 This is likely to reflect the transient nature of the phenomenon, the complex underlying pathophysiological mechanisms involved, and overlap of acute neurology, cardiology, and psychiatric disciplines (see table 1).

Table 1

Causes of syncope*

The goal of the emergency physician, after exclusion of a life threatening condition, is to decide whether inpatient or outpatient management is warranted based on subsequent risk of mortality and morbidity. Accurate early diagnosis of a true seizure disorder will facilitate appropriate disposal and enable important future decisions to be made regarding medical management, work, and driving.

Often there is little history and with few objective clinical signs, establishing a working diagnosis in the A&E department can be difficult. Furthermore, clinical evaluation and routine investigation in the A&E department may be inadequately sensitive to separate seizure from syncope. A diagnostic accuracy of 45% has been suggested using history and examination alone in predicting the cause of syncope.2 Electroencephalography and tilt testing for example may eventually be required to make a definitive diagnosis. In 1978 a raised serum prolactin concentration was found to be a useful marker of a generalised tonic clonic event if measured within an hour of the event.3 Since then its use as a biochemical diagnostic test after a true convulsive episode, pseudoseizure, and non-convulsive syncope has been subject to much interesting debate.

This study seeks to investigate the following: In patients presenting with syncope does a serum prolactin concentration of greater than three times the baseline accurately predict the occurrence of a GTCS?

PHYSIOLOGY OF PROLACTIN SECRETION

Human prolactin is synthesised by the anterior pituitary, secreted in an episodic manner, and is unique in that it is under tonic hypothalamic inhibition by dopamine (prolactin inhibiting factor). Circadian fluctuations occur with peak levels during the middle to end of the night with little change during the waking state.

Physiological causes of hyperprolactinaemia include pregnancy, lactation, sexual arousal, and stress. Destructive lesions of the hypothalamus, pituitary tumours (prolactinoma), and use of dopamine antagonists (phenothiazines, methyldopa, and morphine) also raise serum prolactin. Dopamine agonists lower serum prolactin (bromocriptine, L-dopa, and apomorphine). Baseline adult concentrations are up to 20 ng/ml in the non-pregnant woman and 10 ng/ml in adult men with a plasma half life of about 20 minutes.4

Units are expressed as micro international units per litre (miu/l) or as nanograms per millilitre (1 ng/ml = 21.2 miu/l). Electrochemical stimulation of the medial basal hypothalamus in animal models increases prolactin release. It is thought that in epilepsy, abnormal electrical activity leads to a suppression of dopamine (synthesised in the arcuate and periventricular nuclei of the medial basal hypothalamus) resulting in a raised serum prolactin.5

METHODS

Search strategy

Medline (1966 to Jan 2003) and EMBASE (1982 to Jan 2003) were searched using the Ovid interface and the following keywords: [{(*serum prolactin/ or prolactin.mp. or serum prolactin level/di) AND (generalised tonic clonic seizure or seizure disorder or seizure or fit$).mp. OR (pseudoseizure or hysteria or hysterical fit).mp. OR (syncope or non convulsive syncope or syncopal attack).mp.} LIMIT to (human and english language)]. The Cochrane Library, Pubmed, and key journals relating to emergency medicine were searched using the same search string in addition to a manual review of bibliographies of identified articles.

Study selection

A framework was used for considering study quality and likelihood of bias.6 (See table 2). Study features included the patient sample, reference diagnosis, and the experimental test itself.

Table 2

Framework for considering study quality and likelihood of bias6

Data extraction

Meta-analysis of studies of diagnostic accuracy is a two stage process. Stage 1 involved derivation of summary statistics for each study—that is, determining the sensitivity (SN), specificity (SP), and likelihood ratios (+LR, −LR). Stage 2 involved calculation of the pooled estimate of the SN, SP, and likelihood ratios. Data analysis and confidence intervals were calculated using CIA (confidence interval analysis) software7 and with reference to Sackett et al.8

RESULTS

Thirteen relevant studies were identified in the search. Only three met the criteria required for diagnostic test assessment. (See comments, table 3 and 4). Serum prolactin appears to be released at the onset of a GTCS reaching a peak increase at 15–25 minutes.3 Values remain significantly increased up to one hour after cessation of fit.9,10,16,20 In the evaluation of the “first fit” patient, a baseline pre-ictal value is likely to be unavailable. Recovery prolactin values 24 hours or more after the seizure closely resemble those in control non-stressed subjects and hence provide an acceptable measure of the baseline value for these patients.21 In the diagnosis of a GTCS video EEG is the accepted gold standard. Tables 5 and 6 summarise diagnostic yield of serum prolactin in differentiating a GTCS from pseudoseizure and syncope.

Table 3

Studies of serum prolactin in GTCS and pseudoseizure disorder. Studies meeting criteria for evaluation

Table 4

Studies of serum prolactin in syncope. Studies included in analysis

Table 5

Estimates of two eligible studies of diagnostic accuracy of serum prolactin concentration (three times the baseline within one hour of collapse) in detecting GTCS from pseudoseizure

Table 6

Estimates of one eligible study of sensitivity and specificity of serum prolactin value (three times baseline within one hour of collapse) in detecting GTCS from non-cardiac non-convulsive syncope

DISCUSSION

Modifiers of the test

Alcohol and medication

Seizures associated with alcohol withdrawal show a rise in serum prolactin to values similar to those seen with other causes of GTCS.20 People with epilepsy stabilised with phenytoin seem to have higher baseline values than non-medicated epileptic subjects.22 Diazepam, commonly used to terminate a seizure does not affect serum prolactin values.23 There are no data relevant to newer anti-epileptic medications, SSRIs, and HRT influence on serum prolactin.

Stress, temperature, and children

Patients presenting with stressful acute medical conditions, which are non-seizure related, for example acute unstable angina, asthma attack, and acute abdomen show modestly raised serum prolactin values when compared with healthy subjects on arrival to the A&E department, which do not appear to rise over a 60 minute period.21 In children normal values are less than 17 ng/ml for 0–1 years and less than 9.4 ng/ml for 2–14 years, with slightly higher values in girls than boys. Temperature by itself does not account for a rise in serum prolactin, and values are therefore normal in febrile children. After a febrile seizure, patients’ show similar rises to non-febrile seizure patients, thus “fit” regardless of fever will result in raised prolactin values.24

Validity of the evidence

Patients in two of the studies13,18 evaluated underwent the diagnostic test in question and the reference gold standard (EEG). In the third study3 the clinical criteria were thought acceptable for defining the “non-target” disorder pseudoseizure group. All studies involved a tertiary care cohort, which may have introduced selection bias, a representative sample would have been a prospective A&E department cohort. The three studies included evaluated patients with the target disorder and disorders commonly confused with the target disorder of interest—that is, pseudoseizure and non-convulsive syncope at a point in the clinical process where the test could conceivably be used. All studies suffered from small sample sizes reflected in the wide confidence intervals.

Clinical application of the test

In the absence of a reliable baseline value, a one off serum prolactin value is unlikely to affect decision making in the A&E department. Patients, however, often suffer months of uncertainty while being investigated for “blackouts” and by taking advantage of the time in the A&E department a sensible prediction can be made as to the presence of a GTCS. Serum prolactin should therefore aid management in the neurology clinic and give the patient the benefit of an earlier diagnosis and perhaps enable earlier start of effective treatment. Furthermore, appropriate interpretation of a post-syncopal value against baseline patient value should avoid a false negative risk and potential for labelling the patient with pseudoseizure.

The turn around time in the A&E department is about one to two hours for a serum prolactin result, similar to that of routine biochemistry. This takes into account the time taken for sampling, transport to the laboratory, centrifugation, booking, and analysis using an Architect (Abbott Laboratories) automated immunoassay analyser at a cost of £5 per sample (personal communication with chief MLSO Northwick Park Hospital). Knowledge of other characteristics of a particular patient affecting their prior probability of GTCS in combination with serum prolactin as part of a diagnostic algorithm should increase the diagnostic accuracy.

The potential of a false positive risk is theoretically possible with patients being inappropriately labelled as “epileptic”. For this reason it is important to emphasise the test result gives a risk and not a disease state.

This study has looked at an arbitrary cut off value of three times the baseline value. The predictive value of a threshold effect has not yet been established.

Suggested improvements

Future studies comparing several diagnostic tests, for example, serum creatinine phosphokinase, serum lactate, or a testing algorithm (in combination with clinical, biochemical and brain imaging studies) in properly designed comparative studies in a prospective emergency department cohort are required to investigate the negative predictive value of serum prolactin.

Practice points

  • When differentiating seizure from other causes of transient loss of consciousness serum prolactin values taken within one hour of the event can be useful

  • Always consider physiological, pathological, and iatrogenic causes of raised serum prolactin when interpreting results

  • A baseline comparison value should be taken, seizure free at a similar time of day, 24 hours later

CONCLUSIONS

A 24 year old patient suffers a single episode of syncope. Clinical evaluation reveals no obvious cause or important sequelae. Is a serum prolactin concentration useful in excluding a seizure disorder?

  • In the absence of pregnancy, lactation, and the use of dopamine antagonistic drugs, a level three times the baseline is highly specific for a convulsive event if taken within one hour after cessation of the syncopal episode.

  • A negative result does not exclude a syncopal episode secondary to a GTCS.

  • Raised values seem to be independent of aetiology of GTCS.

RECOMMENDATIONS

  • A&E department physicians should measure the serum prolactin to facilitate eventual diagnosis and medical management.

  • Further research is required ideally in a large prospective A&E department cohort of patients presenting with syncope to fully demonstrate the potential of serum prolactin measurement

Contributors

SA carried out the literature search, collected, processed and analysed the data, and wrote the paper. MB initiated the original idea and edited the paper. SA acts as guarantor.

Footnotes

  • Funding: none.

  • Competing interests: none declared.

REFERENCES


Free sample
This recent issue is free to all users to allow everyone the opportunity to see the full scope and typical content of EMJ.
View free sample issue >>

Don't forget to sign up for content alerts so you keep up to date with all the articles as they are published.