It has been suggested that circulating matrix metalloproteinases (MMP) and tissue inhibitors of matrix metalloproteinases (TIMP) may be useful diagnostic and prognostic indicators in malignant and nonmalignant diseases.1 Numerous reports seem to confirm that assumption although it is unknown whether increased MMP concentration are based on the initial step of tissue damage or the following repairing process.2 There are also contrasting results when similar patients were studied, i.e., increased plasma MMP-9 concentrations but unchanged serum concentrations were found in patients with colon carcinoma.3, 4 We believe that in addition to the analytical and clinical reasons to explain these differences between various studies, more attention should be given to the preanalytical conditions of sampling. In this respect, we read with great interest a recent article published in this journal.5 Gianelli et al.5 found no difference in serum concentrations of MMP-9 and TIMP-1 between healthy women and patients with breast cancer, but decreased MMP-9 and increased TIMP-1 after surgery in comparison with initial concentrations. The authors concluded that these changes might serve as prognostic indices in the follow-up of breast cancer patients. It should be remembered that, however, that using serum material as samples for the determination of those analytes, the authors had rather inadequately considered the impact of blood collection as an important preanalytical determinant of MMP and TIMP results. There is rising evidence that blood sampling and handling markedly determine the concentrations of circulating MMP and TIMP.6, 7, 8, 9, 10, 11, 12 Thus, to interpret that data correctly and to avoid wrong expectations in future studies using these markers, we want to draw the attention of interested clinicians to these facts that were particularly discussed in analytical journals.6, 7, 8, 12 Although the pitfalls in the measurement of circulating growth factors and cytokines for diagnostic purposes have become generally accepted,13 similar evidences for MMP and TIMP have been widely ignored.
Results of own experiments (summarized in Fig. 1) demonstrate the differences of MMP-9 and TIMP-1 in serum and plasma as the analytes come into question. Briefly, for MMP-9 measurements, blood samples from 12 healthy volunteers were collected in Vacutainer tubes (Becton Dickinson, Franklin Lakes, NJ) containing a clot activator for serum preparations or lithium heparinate, dipotassium EDTA or sodium citrate for plasma preparations. Corresponding prepared plastic tubes (Monovettes; Sarstedt, Nürtingen, Germany) were used for blood sampling for TIMP-1 measurements. For the MMP-9 experiments, the tubes were either centrifuged (1,600g, 15 min) immediately (considered as t0) or after 0.5 or 2 hr storage at room temperature and after 24 hr storage at 4°C, respectively. Within 5 days of sampling while the samples were stored at 4°C MMP-9 measurements were carried out using our own enzyme immunoassay described previously.14 Briefly, samples (and standards) were incubated overnight at room temperature in microplate wells coated with our own polyclonal sheep anti-human MMP-9 raised against MMP-9 purified from U937 cell supernatant. After incubation and washing steps, sheep anti-MMP-9 (The Binding Site, Birmingham, UK) was added for 2 hr at room temperature on a rotating platform. After washing, plates were reincubated for 30 min at 20°C with peroxydase-labeled rabbit anti-sheep antibodies (DakoCytomation, Glostrup, Denmark) before adding the substrate 1,2-phenylenediamine for 10 min and measuring the absorbance at 492 nm. For the TIMP-1 experiments, the samples were prepared within 30 min after venipuncture and stored at −80°C until assayed using a commercially assay for TIMP-1 measurements (Biotrak; Amersham, Little Chalfont, UK). Statistical analyses were carried out by ANOVA using the Fisher test. Values (p < 0.05) were considered as statistically significant.
Figure 1a shows that MMP-9 concentration depends on the time between venipuncture and centrifugation as well as on the type of sample preparation. The MMP-9 concentration in serum was about 7–10 times higher than in plasma. Similar results were shown for TIMP-1 although the effect of time between blood sampling and centrifugation was not investigated in detail. The TIMP-1 concentrations were about 5–7 times higher in serum than in plasma (Fig. 1b). Because platelets and leukocytes contain high concentrations of MMP-9 and TIMP-1,15, 16 the observed differences between plasma and serum are attributed to the different release of these analytes from blood cells during platelet activation or sampling process.11, 12 Thus, it can be assumed that serum MMP-9 and TIMP-1 as measured by Giannelli et al.5 are possible misleading markers, because they obviously reflect a high nonspecific background by the release from blood cells and are not in direct relationship to cancer. In addition, Giannelli et al.5 did not describe the time between venipuncture and centrifugation and whether they used tubes available commercially with or without clot activator for serum preparation. Blood sampling with clot activator results in serum samples with higher MMP-9 concentrations.9 As shown previously in patients with renal cell carcinoma, the diagnostic performance of MMP-9 is remarkably influenced by the sample used for measuring MMP-9.9
In conclusion, blood sampling and handling have to be considered as notable preanalytical factors to investigate the concentrations of MMP and TIMP in the peripheral blood and pathological processes in tissue/organs and to use these analytes as diagnostic or prognostic markers. Serum samples should definitely be excluded for these measurements. The use of blood samples collected with sodium citrate was suggested recently as material of choice for measurement of MMP-9 and MMP-2 to optimize the diagnostic validity in peripheral blood.11, 12, 17 So far, no such systematic studies for TIMP were carried out.