What a CMP Measures, and What It Cannot Tell You

A comprehensive metabolic panel is fourteen chemistry measurements on a single serum or plasma sample: eight that describe fluid, electrolyte, and acid-base status plus kidney filtration, and six that describe liver synthesis, hepatocyte integrity, and biliary flow. It is cheap, fast, and highly standardized, which makes it the best repeated-measures substrate most people already have. It is not a cancer screen, and it does not include vitamin D, lipids, thyroid hormones, or a blood count. A single out-of-range value on a fourteen-analyte panel is common enough by chance alone that the result you should care about is a trend. A lone flag is not that result. Interpretation of an abnormal panel belongs with a clinician who knows your history.

The fourteen analytes and what each measurement represents

The electrolyte and renal group is sodium, potassium, chloride, bicarbonate (usually reported as CO2 or total carbon dioxide), glucose, blood urea nitrogen (BUN), creatinine, and calcium. Sodium and chloride are measured on ion-selective electrodes and are tightly regulated, with typical adult intervals near 135–145 mmol/L and 98–107 mmol/L. Bicarbonate is a stand-in for metabolic acid-base status and runs roughly 22–29 mmol/L. BUN (about 7–20 mg/dL) reflects protein intake, hepatic urea synthesis, and renal clearance together, so it moves with hydration and diet more than creatinine does.

Creatinine is the workhorse. It is produced from muscle creatine at a rate proportional to muscle mass and cleared mostly by glomerular filtration, which is why a lean person and a heavily muscled person with identical kidney function will have different creatinine values. Modern labs use assays traceable to isotope-dilution mass spectrometry (IDMS). Confirm this on the report, because older non-standardized assays read high by roughly 5–20 percent and break the eGFR equations.

The hepatic group has these analytes.

• total protein
• albumin
• alkaline phosphatase (ALP)
• alanine aminotransferase (ALT)
• aspartate aminotransferase (AST)
• total bilirubin

ALT is relatively liver-specific. AST is also abundant in skeletal muscle and red cells, so a hard training session or a hemolyzed sample raises it without any hepatic event. ALP comes from liver, bone, intestine, and placenta, and an isolated elevation without a fractionation or a gamma-glutamyl transferase (GGT) measurement is uninterpretable as to source. Albumin (roughly 3.5–5.0 g/dL) is a slow variable, with a half-life around three weeks, so it reports on the past month rather than the past day.

Derived quantities worth computing yourself

Most lab portals report the raw analytes and one or two derived values. The rest are trivial to compute and carry more information than any single number.

The anion gap is Na - (Cl + HCO3), with a reference range near 3–11 mmol/L on current ion-selective electrode platforms, lower than the 8–12 taught from older methods. Globulin is total protein - albumin, and the albumin-to-globulin ratio is a cheap summary of synthetic function versus immunoglobulin load. Albumin-corrected calcium is measured Ca + 0.8 × (4.0 - albumin) in mg/dL and g/dL, which matters because roughly 40 percent of serum calcium is albumin-bound.

For filtration, use the 2021 race-free CKD-EPI creatinine equation:

kappa = 0.7 (female), 0.9 (male)
alpha = -0.241 (female), -0.302 (male)
eGFR = 142 * min(Scr/kappa, 1)**alpha
          * max(Scr/kappa, 1)**-1.200
          * 0.9938**age
          * (1.012 if female else 1.0)

Report it as mL/min/1.73 m², and treat values above 60 as imprecise by construction. If you have a complete blood count (CBC) from the same draw, FIB-4 is (age × AST) / (platelets × sqrt(ALT)) with platelets in 10⁹/L, a score designed for population-level stratification of hepatic fibrosis risk rather than individual diagnosis.

Preanalytical error causes most surprising results

Before you believe an outlier, check whether the sample was handled in a way that could produce it. In our experience this explains the majority of one-off abnormalities in otherwise stable people.

Hemolysis is the single largest offender. Intracellular potassium is roughly twenty-five times plasma potassium, so even slight red cell lysis from a small-gauge needle, vigorous mixing, or pneumatic tube transport raises potassium, AST, and lactate dehydrogenase. Good labs report a hemolysis index. Ask for it. Fist clenching during venipuncture can raise potassium by several tenths of a mmol/L on its own. A tourniquet left on beyond a minute concentrates protein-bound analytes, raising albumin, total protein, and calcium by a few percent through hemoconcentration. Standing for fifteen minutes before a draw does the same thing.

Order-of-draw errors are recognizable by pattern: carryover from an EDTA (lavender) tube chelates calcium and zinc, so you see implausibly low calcium and ALP alongside a high potassium in the same sample. Glucose is the most time-sensitive analyte, falling on the order of 5–7 percent per hour at room temperature in a plain tube as red cells continue glycolysis, which is why sodium fluoride tubes or rapid centrifugation matter if glucose is the point of the draw.

Deciding when a change is real

A CMP becomes interesting when you have five or ten of them. To decide whether a delta is signal, use the reference change value, RCV = 1.41 × Z × sqrt(CV_A² + CV_I²). Here CV_A is the analytical imprecision of your lab’s platform, which you can ask for or estimate from proficiency data. CV_I is within-subject biological variation. The two extremes are instructive. Sodium has a within-subject CV under one percent, so a 4 mmol/L move is large relative to your own setpoint even though it sits inside the population interval. ALT has a within-subject CV in the range of 20 to 25 percent, so a doubling between two draws can occur with no underlying change at all.

This is the argument for treating population reference intervals as a weak prior and your own history as the strong one. A creatinine of 1.15 mg/dL is unremarkable in the population and notable if your last eight values clustered at 0.85.

What a CMP says about cancer, which is less than people hope

A CMP does not screen for cancer. Some malignancies eventually perturb it, through hypercalcemia, elevated ALP with bone or biliary involvement, rising bilirubin with biliary obstruction, or falling albumin with systemic inflammation, but these appear late and have many benign causes. The general-purpose blood markers that have been studied specifically for early detection perform poorly enough in screening populations that systematic review has repeatedly found single markers inadequate for that purpose, with sensitivity at acceptable specificity far below what a screening test requires.¹ Current work in this area has moved toward circulating tumor DNA, proteomic panels, and multi-analyte combinations rather than routine chemistry.²

There is a second trap worth understanding if you ever add tumor markers to a chemistry panel out of curiosity. Renal impairment, hepatic dysfunction, and inflammation raise many conventional tumor marker concentrations in people without any neoplasm, so an abnormal CMP and an abnormal marker in the same draw are often two readouts of the same non-malignant physiology.³ Multimarker panels combining several analytes have shown better discrimination than any single one in specific research settings, which is a statement about study cohorts rather than about screening yourself.⁴

Getting the data into a form you can work with

Ask for results as HL7 or FHIR Observation resources with LOINC codes rather than a PDF. LOINC identifies the analyte, specimen, and method, so sodium in serum is 2951-2 regardless of which lab produced it, and codes are what make a decade of draws from four different providers joinable. Store the reference interval and the performing lab with every value, because intervals differ across platforms and a change in lab will otherwise look like a change in you. If you only have PDFs, parse them into a long-format table and normalize units at ingest (creatinine in mg/dL versus µmol/L is the usual break). Give the table these columns.

• loinc
• analyte
• value
• unit
• ref_low
• ref_high
• collected_at
• lab

Plot each analyte against time with the reference band shaded, and you will learn more in ten minutes than from any single report.

Questions people also ask

Can I do a CMP without fasting? Yes, for thirteen of the fourteen analytes. Glucose is the exception and requires an 8 to 12 hour fast to be interpretable against fasting reference ranges. A non-fasting glucose is still a valid measurement, just of a different quantity.

What is the difference between a CMP and a CBC? A CMP measures dissolved chemistry in serum or plasma. A CBC counts and characterizes the cells: red cells, white cells with differential, platelets, and indices like mean corpuscular volume. They are different tubes and different instruments, which is why they are usually ordered together.

Will a CMP show liver problems? It measures four liver-related quantities: ALT and AST for hepatocyte leakage, ALP and bilirubin for biliary flow, plus albumin for synthetic capacity. It is sensitive to many hepatic processes and specific to none, and a normal panel does not exclude significant liver disease, including established fibrosis.

Does a CMP test for vitamin D? No. 25-hydroxyvitamin D is a separate assay, usually by immunoassay or liquid chromatography–tandem mass spectrometry, and must be ordered on its own.

How long do results take? Chemistry panels run in minutes on an automated analyzer. The delay is logistics: same-day for hospital labs, typically one to two business days for outpatient draws sent to a central lab.

Should I be worried if my CMP is abnormal? With fourteen analytes and reference intervals covering the central 95 percent of a reference population, roughly half of healthy people will have at least one value outside range on any given panel. Repeat the measurement, check for the preanalytical causes above, and bring persistent or large deviations to a clinician rather than interpreting them alone.

Footnotes

1. Sabrina Hundt, Ulrike Haug, Hermann Brenner. Blood Markers for Early Detection of Colorectal Cancer: A Systematic Review. Cancer Epidemiology, Biomarkers & Prevention, 2007. https://doi.org/10.1158/1055-9965.epi-06-0994 ↩

2. Samir M. Hanash, Christina S. Baik, Olli Kallioniemi. Emerging molecular biomarkers—blood-based strategies to detect and monitor cancer. Nature Reviews Clinical Oncology, 2011. https://doi.org/10.1038/nrclinonc.2010.220 ↩

3. Jaume Trapé, Esther Fernández-Galán, Josep Maria Auge, et al. Factors influencing blood tumor marker concentrations in the absence of neoplasia. Tumor Biology, 2024. https://doi.org/10.3233/tub-220023 ↩

4. Z YURKOVETSKY, S TAASAN, S SKATES, et al. Development of multimarker panel for early detection of endometrial cancer. High diagnostic power of prolactin. Gynecologic Oncology, 2007. https://doi.org/10.1016/j.ygyno.2007.05.041 ↩