The Light Criteria

The Beginning and Why they are Useful 40

KEYWORDS
 Light criteria  Pleural effusion  Transudative effusion

KEY POINTS
 The Light criteria serve as a good starting point in the separation of transudates from exudates.
 The Light criteria misclassify about 25% of transudates as exudates, and most of these patients are
on diuretics.
 If a patient is thought likely to have a disease that produces a transudative pleural effusion but the
Light criteria suggest an exudate by only a small margin, the serum–pleural fluid protein gradient
should be examined.
 If this is greater than 3.1 gm/dL, the patient in all probability has a transudative effusion.
 If the gradient is less than 3.1 gm/dL, either the NT-pro-BNP level in the pleural fluid or the serum–
pleural fluid albumin gradient can be measured.
 Either an NT-pro-BNP greater than 1300 pg/mL or an albumin gradient greater than 1.2 gm/dL indi-
cate that the effusion is a transudate.

It has been 40 years since I published the article1 and, for the most part, neither he nor anybody
describing what came to be known as the Light else had a scientific answer.
criteria. I thought that it might be appropriate to It was at this time that additional measurements
begin this article by detailing how that article were first being made on blood, such as the lactic
came about. dehydrogenase (LDH), aspartate aminotrans-
ferase (AST), and alanine aminotransferase (ALT).
THE DEVELOPMENT OF THE LIGHT CRITERIA At about the same time, blood gas machines
became available that would allow the accurate
When I was an intern in medicine at Johns Hopkins measurement of pH, PCO2, and PO2 of body fluids.
Hospital in Baltimore, Maryland, in 1968 to 1969, I theorized that some of these new measurements
there was a period when a large percentage of might be useful in the differential diagnosis of
my patients had a pleural effusion. The chief resi- pleural effusions. After doing a literature review,
dent, Dr Richard Winterbauer, made rounds I developed 2 hypotheses. The first was that the
around midnight and always asked me what the pH of pleural fluid would be lower in tuberculous
thoracentesis revealed. At that time, we routinely pleural effusions than in other exudative pleural
measured the cell count and differential, glucose, effusions. The basis for this hypothesis was an
and protein, and performed smears and cultures article in the Scandinavian Journal of Respiratory
on the pleural fluid. I asked Dr Winterbauer the Disease that purported to show this.2 My second
chestmed.theclinics.com

significance of the various pleural fluid findings hypothesis was that LDH isoenzymes would be

Disclosures: The author has nothing to disclose related to this publication.

Clin Chest Med 34 (2013) 21–26

useful in the differential diagnosis of exudative When I examined my plots, it was obvious that
pleural effusions. To get the absolute value of the no single value of any of these measurements
LDH isoenzymes, I needed to have the total LDH correctly identified all transudates and exudates.
in the pleural fluid and the serum. A previous study If the cutoff was made high enough that all transu-
on pleural fluid LDH concluded that the pleural dates were below the cutoff level, then some
fluid LDH was increased in malignant pleural effu- exudates would be classified as transudates. My
sions compared with other pleural effusions.3 objective at that time was to identify all exudates
I submitted a proposal to the Institutional Review correctly. Therefore I elected to make the cutoff
Board and received their approval. The blood points such that no transudates were above the
gas machine was in the pulmonary function labo- line. I noticed that, when I did this, some exudates
ratory and I could measure all of the pleural fluid were in the transudative range for each of the
pH myself. The clinical laboratory measured the measurements. However, I also noticed that, if
protein, LDH, and glucose in the serum and pleural I used 3 different cutoff levels such that no transu-
fluid without charge. However, I did have to come dates were above the cutoff line, I could identify
up with some funds to pay for the LDH isoen- almost all transudates and exudates correctly.
zymes. I received a small grant from Johns Hop- The 3 cutoff points that I found were a protein ratio
kins Hospital to fund this. greater than 0.5, an LDH ratio greater than 0.6, and
To get called when patients with pleural effu- an absolute pleural fluid LDH greater than two-
sions were admitted, I made a deal with my fellow thirds of the upper normal limit for serum. An
interns and residents. If they would call me when exudative effusion met at least 1 of these 3 criteria,
they did a thoracentesis, I would do the cell count whereas a transudative effusion met none.
and differential on the pleural fluid; duties for which I presented these data to the alumni and they
they would normally be responsible. I quickly did not seem particularly impressed. I also sub-
found out that, with this arrangement, I got called mitted an abstract on the separation of transu-
often in the middle of the night about pleural dates by the criteria listed earlier to the American
effusions. College of Physicians in 1972.6 It was accepted
One of the first patients I studied was a young for an oral presentation in Atlantic City. This oral
man with an exudative lymphocytic effusion. His presentation was the only one that I ever partici-
pleural fluid pH was 7.40. The patient turned out pated in where the audience graded the contents
to have caseating granulomas on the needle of the presentation. I got, at most, average marks;
biopsy of his pleura; so much for the first hypoth- certainly nothing to suggest that these cutoff levels
esis. Soon thereafter, another patient had a pleural would still be in use 40 years later. Nevertheless,
fluid pH of 6.95. The pleural fluid was clear yellow I wrote the article and submitted it to the Annals
and the pleural fluid glucose was not reduced. of Internal Medicine. It was accepted with minimal
However, the pleural fluid grew Streptococcus revisions.1
pneumoniae and the patient eventually developed There are several lessons to be learned from my
a frank pneumococcal empyema. This case was experience in developing the Light criteria. First, if
the first to suggest that a low pleural fluid pH might you want people to cooperate with you on your
be an indicator of a complicated parapneumonic research, you need to make it worthwhile for
effusion.4 them. In this case, I did some of the work that
I subsequently studied more than 150 pleural they would otherwise have to do. Second, al-
effusions in a 2-year period. I submitted an though research is best done when it is hypothesis
abstract of my preliminary findings to the American driven, it is worthwhile to look at your data to
Thoracic Society for their annual meeting in 1971. determine whether there are other interesting find-
The abstract was rejected. I was devastated. ings. Third, if you initially submit your work and it is
In early 1971, when I was a pulmonary fellow, not particularly well received, do not give up.
Johns Hopkins had a reunion for some of its Remember that the first abstract on the Light
alumni. My mentor, Dr Wilmot C. Ball, Jr, sug- criteria was turned down.
gested that I present something on the pleural fluid
that I had been studying. At that time, transudates WHY THE LIGHT CRITERIA ARE STILL USEFUL
and exudates were usually separated by using
a protein level of 3.0 gm/dL.5 I elected to see The first reference to use the name the Light criteria
how this would work on my set of pleural effusions. that I am aware of was published in 1989.7 Since the
On a rainy, sleety Sunday in Baltimore, Maryland, original publication in 1972, there have been many
I spent several hours with a pencil and graph paper studies comparing other measurements with the
plotting protein levels, LDH levels, and ratios of Light criteria for the separation of transudates and
protein and LDH in the serum and pleural fluid. exudates, but, in general, the Light criteria have
 Light Criteria 23

been proved to be better than anything else. I am pleural fluid to the serum protein level.12 Choles-
amazed that, after 40 years, the Light criteria are terol measurements provide no additional informa-
still being used. tion to the protein ratio. Other proposed measures
Pleural effusions have classically been divided have included the pleural fluid/serum bilirubin
into transudates and exudates. By definition, ratio,13 the pleural fluid viscosity,14 the level of
a transudative pleural effusion develops when oxidative stress markers,15 the level of soluble
the systemic factors influencing the formation or leukocyte selectin,16 the level of various cyto-
absorption of pleural fluid are altered so that kines,17 the level of uric acid,18 and the pleural
pleural fluid accumulates. The pleural fluid is called fluid/serum cholinesterase ratio.19 Subsequent
a transudate. The permeability of the capillaries to articles20,21 have shown that none of these other
proteins is normal in the area where the fluid is measures are superior to the Light criteria in sepa-
formed. Examples of conditions producing transu- rating transudates from exudates.
dative pleural effusions are left ventricular failure When the original article1 was published, the
producing increased pulmonary interstitial fluid Light criteria identified all transudates and ex-
and a resulting pleural effusion, ascites caused udates accurately. However, subsequent studies
by cirrhosis with movement of fluid through the have shown that the Light criteria classify almost
diaphragm, and decreased serum oncotic pres- all exudates correctly, but falsely classify about
sure with hypoproteinemia.8 25% of transudates as exudates. In a study22 of
In contrast, an exudative pleural effusion 249 patients in which there were 185 exudates
develops when the pleural surfaces or the capil- and 64 transudates, the Light criteria correctly
laries in the location where the fluid originates identified 99.5% of exudates but only 75% of tran-
are altered such that fluid accumulates. The most sudates. In a recent article,23 the Light criteria
common causes of exudative pleural effusions falsely classified 107 of 364 (29%) transudates
are pleural malignancy, parapneumonic effusions, caused by heart failure and 18 of 102 (18%) transu-
and pulmonary embolism. dates caused by hepatic hydrothorax.
Why is it important to differentiate transudates Transudates that are falsely classified as ex-
from exudates? If a patient has a transudative udates usually make exudative criteria by only
pleural effusion, then it is only necessary to treat a slight amount. In an article23 involving 107 falsely
the cause of the effusion, such as heart failure or classified transudates, the mean pleural fluid (PF)/
cirrhosis. However, if it is an exudative effusion, serum protein ratio was 0.51, the mean PF/serum
more investigation is indicated to identify the local LDH was 0.63, and the mean pleural fluid LDH was
problem that is causing the pleural effusion. 0.34 the upper normal limit for serum for the transu-
Why use 2 different measurements to separate dates that were falsely classified. Most transudates
transudates from exudates? The pleural fluid/ are falsely classified by only 1 of the 3 elements of
serum protein ratio is an indication of the perme- the Light criteria.23 Most of the patients with transu-
ability of the capillaries in the area where the pleural dates who are falsely classified either are receiving
fluid was formed. In contrast, the pleural fluid LDH diuretics or have a pleural fluid red blood cell
level is an indication of the degree of inflammation (RBC) count greater than 10,000/mm3.24 Romero-
in the pleural space. Therefore, they measure Candeira and colleagues25 performed 3 thoracente-
different things. However, there is no doubt that ses at 48-hour intervals in 15 patients with pleural
they are related because inflammation in the pleural effusions caused by heart failure after they were
space increases the permeability of the capillaries started on diuretics. They reported that the mean
in the pleura. pleural fluid protein increased from 2.3 to
Why are the Light criteria still useful after 3.3 gm/dL and the mean pleural fluid LDH
40 years? I think that there are several reasons increased from 177 to 288 IU/L. After diuresis, the
for its popularity and its usefulness, including (1) Light criteria would have classified most of
it is simple and easy to remember, (2) its measure- the effusions as exudates.25
ment is readily available, (3) it is accurate. In the original article,1 almost all transudates
Since the Light criteria were first presented, were classified correctly. What is the explanation
there have been many articles proposing alterna- for the observation that currently about 25% of
tives. One of the measures proposed was the level transudates are falsely classified? I think that there
of cholesterol in the pleural fluid,9–11 or the ratio of are probably 2 reasons for this discrepancy. First,
the pleural fluid to the serum cholesterol.11 The in the original series, almost all the thoracenteses
cholesterol measurement is unlikely to be superior were done on patients who had just been admitted
to the Light criteria because the pleural fluid to the hospital and were not on diuretics. Sec-
cholesterol level can be accurately predicted ond, the diuretics are more powerful now than
from the serum cholesterol and the ratio of the those that were available in 1972, namely
24 Light

hydrochlorothiazide and the mercurial diuretics (ie, a patient who could have a transudative effusion
mercury chloride; now rarely used). but who meets exudative criteria via the Light
An alternative approach for the identification criteria by a small margin, I first look at the
of transudates and exudates is to select the serum–pleural fluid protein gradient. If this is
cutoff levels that correctly identify the highest greater than 3.1 gm/dL, I conclude that the patient
percentage of patients. Heffner and colleagues26 has a transudative effusion. I look at the protein
analyzed data from 8 studies with a total of gradient first because it is already available from
1448 patients and concluded that the best cutoff the Light criteria. If this gradient is less than
levels for the different pleural fluid tests were 3.1 gm/dL, I consider measuring the albumin
protein ratio 0.5, LDH ratio 0.45, and LDH 0.45 of gradient. If the patient may have a hepatic hydro-
the upper limits of normal for serum. The problem thorax, I measure the albumin ratio. As an alterna-
with this approach is that some effusions will be tive, measurement of the serum or pleural fluid
misclassified but it is impossible to know whether N-terminal probrain natriuretic peptide (NT-pro-
they are misclassified wrongly as a transudate or BNP) can be performed to determine whether the
an exudate. I still prefer the Light criteria because pleural effusion is caused by CHF (discussed than).
I want to definitely identify all exudative effusions. Most transudates are caused by CHF. It is prefer-
able to establish the diagnosis of CHF directly. It
TRANSUDATIVE EFFUSIONS MISCLASSIFIED seems that the diagnosis of CHF as a cause of the
BY THE LIGHT CRITERIA pleural effusion can be made with measurement
of the NT-pro-BNP in the pleural fluid or the serum.
When the Light criteria are used, how are those When the ventricles are subjected to increased
transudative pleural effusions identified that are pressure or volume, NT-pro-BNP and BNP are
misclassified? If a patient is clinically suspected released into the circulation.27 The biologically
of having a transudative effusion but exudative active BNP and the larger NT-pro-BNP are released
criteria are met by a small margin (protein ratio in equimolar amounts into the circulation.27
between 0.5 and 0.65, LDH ratio between 0.6 and Porcel and colleagues28 first showed that the
1.0, pleural fluid LDH between two-thirds and the pleural fluid levels of NT-pro-BNP are increased in
upper normal limit for serum), attempts should be patients with heart failure. These researchers
made to determine whether the patient really has measured NT-pro-BNP levels in 117 pleural fluid
a transudative effusion. The 2 main measures that samples with the following diagnosis: CHF, n 5 44;
have been proposed to identify these patients are malignancy, n 5 35; tuberculous pleuritis, n 5 20;
a serum–pleural fluid albumin gradient greater hepatic hydrothorax, n 5 10; and miscellaneous,
than 1.2 gm/dL or a serum–pleural fluid protein n 5 18. The mean pleural fluid NT-pro-BNP level in
gradient greater than 3.1 gm/dL. Romero- the patients with CHF was 6931 pg/mL, which was
Candeira and colleagues22 studied 64 patients significantly higher than the 551 pg/mL in patients
with transudative pleural effusions and reported with hepatic hydrothorax and the 292 pg/mL in
that the Light criteria identified 75% correctly, the patients with exudative pleural effusions.28 When
serum–pleural fluid albumin gradient identified a cutoff level of 1500 pg/mL was used, the sensitivity
86% correctly, and the serum–pleural fluid protein was 91% and the specificity was 93% for the diag-
gradient identified 91% correctly. In a recent nosis of CHF.28 We compared the pleural fluid NT-
article,23 107 of 364 transudates (29%) caused by pro-BNP levels in 10 patients each with effusions
congestive heart failure (CHF) were misclassified caused by CHF, pulmonary embolism, postcoro-
as exudates. In these 107 instances, a serum– nary artery bypass graft surgery, and malignancy.29
pleural fluid protein gradient greater than 3.1 gm/ All the patients with CHF had NT-pro-BNP levels
dL identified 55% correctly, whereas a serum– greater than 1500 pg/mL, whereas none of the other
pleural fluid albumin gradient greater than 1.2 gm/ patients had NT-pro-BNP levels this high. There
dL, which was only performed in 36 patients, iden- have been several subsequent articles evaluating
tified 83% correctly.23 In the same report, 18 of the accuracy of NT-pro-BNP in making the diag-
102 transudates (18%) caused by hepatic hydro- nosis of pleural effusion caused by heart failure. In
thorax were misclassified with the Light criteria. a meta-analysis of 10 studies with a total of
The protein gradient and the albumin gradient iden- 1120 patients, the pooled sensitivity and specificity
tified 61% and 62% of these misclassified transu- were 94% and 94% respectively.30
dates correctly. An albumin ratio less than There is a close relationship between the levels
0.6 correctly identified 77% of these transudates of NT-pro-BNP in the pleural fluid and serum.
caused by hepatic hydrothorax. Han and colleagues31 measured the NT-pro-BNP
My recommendations in view of the 2 studies levels in 240 patients and reported that the corre-
discussed earlier are as follows. When evaluating lation coefficient between the pleural and serum
 Light Criteria 25

NT-pro-BNP was 0.928. In a second study, Kolditz 2. Holten K. Diagnostic value of some biochemical
and colleagues32 measured the serum and pleural pleural fluid examinations. Scand J Respir Dis Suppl
fluid NT-pro-BNP levels in 93 patients, including 1968;63:121–5.
25 with CHF. They confirmed the results of the 3. Wroblewski F, Wroblewski R. The clinical significant
study mentioned earlier in that the levels of serum of lactic dehydrogenase activity of serous effusions.
and pleural fluid NT-pro-BNP were closely corre- Ann Intern Med 1958;48:813–22.
lated (r2 5 0.90). In addition, in each of the two 4. Light RW, MacGregor MI, Ball WC Jr, et al. Diag-
studies,31,32 the values in the pleural fluid and the nostic significance of pleural fluid pH and PCO2.
serum were almost identical. From these latter Chest 1973;64:591–6.
two studies, it seems that measurement of the 5. Carr DT, Power MH. Clinical value of measurements
pleural fluid NT-pro-BNP levels provides no of protein in pleural fluid. N Engl J Med 1958;259:
additional information beyond the serum 926–7.
measurements. 6. Light RW, MacGregor MI, Ball WC Jr, et al. Pleural
The pleural fluid NT-pro-BNP is also superior to fluid lactic acid dehydrogenases and protein
the BNP and the protein gradient in identifying content. Ann Intern Med 1972;76:880.
patients with heart failure who meet the Light 7. Scheurich JW, Keuer SP, Graham DY. Pleural effu-
criteria for exudates.33 In a study of 20 patients sion: comparison of clinical judgment and Light’s
with heart failure who met the Light criteria for criteria in determining the cause. South Med J
exudates, 18 had NT-pro-BNP levels greater than 1989;82:1487–91.
1300 pg/mL, 16 had NT-pro-BNP levels greater 8. Light RW. Pleural diseases. 5th edition. Baltimore
than 1500 pg/mL, but only 10 had a serum–pleural (MD): Lippincott Williams and Wilkins; 2007.
fluid protein gradient greater than 3.1 gm/dL. 9. Hamm H, Brohan U, Bohmer R, et al. Cholesterol in
The serum or pleural fluid BNP and NT-pro-BNP pleural effusions: a diagnostic aid. Chest 1987;92:
cannot be used interchangeably in the diagnosis of 296–302.
pleural effusions caused by CHF.34 The BNP levels 10. Valdés L, Pose A, Suarez J, et al. Cholesterol:
are only about 10% of the NT-pro-BNP levels. a useful parameter for distinguishing between
There is not a close correlation between the BNP pleural exudates and transudates. Chest 1991;99:
levels and the NT-pro-BNP levels (r 5 0.78).33,34 1097–102.
Moreover, the diagnostic usefulness of the 11. Costa M, Quiroga T, Cruz E. Measurement of pleural
NT-pro-BNP in making the diagnosis of heart fluid cholesterol and lactate dehydrogenase. A
failure is superior to that of the BNP.33,35 simple and accurate set of indicators for separating
exudates from transudates. Chest 1995;108:1260–3.
12. Vaz MA, Teixeira LR, Vargas FS, et al. Relationship
SUMMARY
between pleural fluid and serum cholesterol levels.
The Light criteria serve as a good starting point in Chest 2001;119:204–10.
the separation of transudates from exudates. The 13. Meisel S, Shamiss A, Thaler M, et al. Pleural fluid to
Light criteria misclassify about 25% of transudates serum bilirubin concentration ratio for the separation
as exudates, and most of these patients are on of transudates from exudates. Chest 1990;98:141–4.
diuretics. If a patient is thought likely to have 14. Yetkin O, Tek I, Kaya A, et al. A simple laboratory
a disease that produces a transudative pleural measurement for discrimination of transudative and
effusion but the Light criteria suggest an exudate exudative pleural effusion: pleural viscosity. Respir
by only a small margin, the serum–pleural fluid Med 2006;100:1286–90.
protein gradient should be examined. If this is 15. Papageorgiou E, Kostikas K, Kiropoulos T, et al.
greater than 3.1 gm/dL, the patient in probably Increased oxidative stress in exudative pleural effu-
has a transudative effusion. If the gradient is less sions: a new marker for the differentiation between
than 3.1 gm/dL, either the NT-pro-BNP level in exudates and transudates? Chest 2005;128:3291–7.
the pleural fluid or the serum–pleural fluid albumin 16. Horvath LL, Gallup RA, Worley BD, et al. Soluble
gradient can be measured. Either an NT-pro-BNP leukocyte selectin in the analysis of pleural effu-
greater than 1300 pg/mL or an albumin gradient sions. Chest 2001;120:362–8.
greater than 1.2 gm/dL indicate that the effusion 17. Alexandrakis MG, Kyriakou D, Alexandraki R, et al.
is a transudate. Pleural interleukin-1beta in differentiating transudates
and exudates: comparative analysis with other
REFERENCES biochemical parameters. Respiration 2002;69:
201–6.
1. Light RW, MacGregor MI, Luchsinger PC, et al. Pleural 18. Uzun K, Vural H, Ozer F, et al. Diagnostic value of
effusions: the diagnostic separation of transudates uric acid to differentiate transudates and exudates.
and exudates. Ann Intern Med 1972;77:507–14. Clin Chem Lab Med 2000;38:661–5.
26 Light

19. Garcia-Pachon E, Padilla-Navas I, Sanchez JF, et al. 27. Pfister R, Schneider CA. Natriuretic peptides BNP
Pleural fluid to serum cholinesterase ratio for the and NT-pro-BNP: established laboratory markers in
separation of transudates and exudates. Chest clinical practice or just perspectives. Clin Chim
1996;110:97–101. Acta 2004;349:25–38.
20. Romero S, Candela A, Martin C, et al. Evaluation of 28. Porcel JM, Vives M, Cao G, et al. Measurement of
different criteria for the separation of pleural transu- pro-brain natriuretic peptide in pleural fluid for the
dates from exudates. Chest 1993;104:399–404. diagnosis of pleural effusions due to heart failure.
21. Burgess LJ, Maritz FJ, Taljaard JJ. Comparative Am J Med 2004;116:417–20.
analysis of the biochemical parameters used to 29. Liao H, Na MJ, Dikensoy O, et al. The diagnostic value
distinguish between pleural transudates and of pleural fluid NT-proBNP levels in patients with
exudates. Chest 1995;107:1604–9. cardiovascular diseases. Respirology 2008;13:53–7.
22. Romero-Candeira S, Hernandez L, Romero- 30. Janda S, Swiston J. Diagnostic accuracy of pleural
Brufao S, et al. Is it meaningful to use biochemical fluid NT-pro-BNP for pleural effusions of cardiac
parameters to discriminate between transudative origin: a systematic review and meta-analysis.
and exudative pleural effusions? Chest 2002;122: BMC Pulm Med 2010;10:58.
1524–9. 31. Han CH, Choi JE, Chung JH. Clinical utility of pleural
23. Bielsa S, Porcel JM, Castellote J, et al. Solving fluid NT-pro brain natriuretic peptide (NT-proBNP) in
the Light’s criteria misclassification rate of car- patients with pleural effusions. Intern Med 2008;47:
diac and hepatic transudates. Respirology 2012; 1669–74.
17:721–6. 32. Kolditz M, Halank M, Schiemanck S, et al. High diag-
24. Porcel JM, Esquerda A, Martinez M, et al. Influence nostic accuracy of NT-proBNP for cardiac origin of
of pleural fluid red blood cell count on the misiden- pleural effusions. Eur Respir J 2006;28:144–50.
tification of transudates. Med Clin (Barc) 2008;131: 33. Porcel JM, Martı́nez-Alonso M, Cao G, et al.
770–2. Biomarkers of heart failure in pleural fluid. Chest
25. Romero-Candeira S, Fernandez C, Martin C, et al. 2009;136:671–7.
Influence of diuretics on the concentration of 34. Porcel JM. Utilization of B-type natriuretic peptide and
proteins and other components of pleural transu- NT-proBNP in the diagnosis of pleural effusions due to
dates in patients with heart failure. Am J Med heart failure. Curr Opin Pulm Med 2011;17:215–9.
2001;110:681–6. 35. Long AC, O’Neal HR Jr, Peng S, et al. Comparison of
26. Heffner JE, Sahn SA, Brown LK. Multilevel likelihood pleural fluid N-terminal pro-brain natriuretic peptide
ratios for identifying exudative pleural effusion. and brain natriuretic-32 peptide levels. Chest
Chest 2002;121:1916–20. 2010;137:1369–74.