All fibrillar collagens (e.g., types I and III) are synthesized as procollagens, which contain additional propeptide extensions at both ends. Prior to collagen fiber formation, the procollagens are converted to collagens. During this process, the propeptide regions of the procollagen molecules are cleaved off in the extracellular space and released into the circulation ^{[1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20]}[81,82,83,84,85,86,158,159,260,261,262,263,264,265,266,267,268,269,270,271]. Since the propeptides are released in equimolar amounts to the collagen molecules formed, the measurement of circulating levels of these propeptides is thought to reflect collagen synthesis. Willumsen et al. summarized this process recently for neo-epitop specific type III collagen propeptides (PRO-C3) measured in serum in a fraction of patients with various solid tumors, highlighting the prognostic value in the context of overall survival (see below Future Research Directions) ^[21][154].

In previous studies of PIIINP in patients with PMF and related neoplasms ^{[1][3][4][5][6][10][11]}[81,83,84,85,86,261,262], the highest serum levels were found in patients with PMF ^{[1][3][4][6][22]}[81,83,84,86,91], in those PV patients transforming into a myelofibrotic stage of the disease ^[1][3][6][81,83,86], and in patients with chronic myelogenous leukemia (CML) associated with bone marrow fibrosis ^[4][6][84,86]. Although these studies suggested serum PIIINP as a useful noninvasive means to monitor the accumulation of interstitial type III collagen in the bone marrow, several factors should be considered when interpreting serum PIIINP levels. First, the radioimmunoassay for the aminoterminal propeptide of type III procollagen (RIA-gnost, Hoechst, FRG), used in the seminal studies on circulating PIIINP in MPNs, did not exclusively detect the authentic propeptide released during the conversion of procollagen to collagen but measured at least three molecular weight variants of the antigens ^[12][13][263,264]. The main antigen in normal serum is smaller than the authentic aminopropeptide liberated during the conversion of type III pN collagen to type III collagen. These lower molecular weight peptides were shown to probably represent degradation products of the propeptide ^[14][15][265,266], including those related to the degradation of newly synthesized type III procollagen or degradation of that portion of the propeptide retained in situ in the bone marrow. However, in rheumatic diseases, the local degradation of pN-collagen in inflamed tissue has not been considered to contribute significantly to elevated serum aminopropeptide levels ^[14][265].

Owing to the heterogeneity of the antigens related to the aminoterminal propeptide of type III procollagen, a next-generation equilibrium RIA for the N-terminal propeptide of human type III procollagen (Farmos Diagnostica, Oulu, Finland) was additionally applied in a series of patients with PMF and related neoplasms ^[22][91]. This assay system did not detect the smaller antigen variants, but only antigen forms equal to or larger than the authentic propeptide. A highly significant correlation was found between the two assay systems, which both discriminated between patients with stable and active disease, the highest propeptide levels being recorded in patients with a syndrome of acute myelofibrosis and those with accelerated disease ^[22][91]. Furthermore, serum propeptide levels correlated significantly with conventional markers of disease activity (leukocyte count and plasma lactic dehydrogenase), which was also observed in longitudinal studies, where a close co-variation between serum PIIINP and these parameters was recorded.

Analysis of the antigen profile of PIIINP by means of gel filtration in patients with different disease activity showed smaller antigen variants to prevail in patients with stable disease, whereas the intact propeptide dominated in patients with acute or transforming disease. This observation indicated that elevated circulating PIIINP levels primarily reflect de novo synthesis of type III collagen and not increased degradation of the propeptide ^[22][91]. Vellenga et al. also found that the antigenicity related to PIIINP was heterogeneous, with at least two main peaks having molecular weights equal to and smaller than the authentic propeptide ^[10][261]. Cytotoxic treatment was accompanied by decreasing serum propeptide levels ^[22][91]. Taken together, these early observations indicated type III collagen metabolism to be closely linked to clonal myeloproliferation ^[6][22][86,91]. Recently, Willumsen et al. introduced an assay that specifically measures propeptide containing the neo-epitope cleavage site generating the true propeptide (PRO-C3) and, therefore, measures the true formation of type III collagen ^[21][23][105,154] (see Future Research Directions).

Second, while an assay for serum PIIINP mainly measures de novo synthesis of type III procollagen at the time of investigation, the degree of bone marrow fibrosis as assessed by a bone marrow biopsy provides evidence about the preceding net collagen deposition ^[5][22][85,91].

This accumulation may be due to both an increased collagen synthesis and/or impaired degradation of bone marrow collagen. Thus, the observed relationship between increasing degrees of bone marrow fibrosis and increasing serum propeptide levels rather reflects an enhanced myeloproliferative activity and hence ongoing stimulation of bone marrow fibroblasts.

Third, circulating type III procollagen-related antigens can be derived from sources other than the bone marrow, including both the spleen and liver, due to myeloid metaplasia and fibroplasia in these organs ^[24][206]. Thus, raised serum PIIINP may theoretically also reflect impaired propeptide metabolism in patients with massive myeloid metaplasia in the liver since serum PIIINP is increased in certain liver diseases ^[12][20][263,271] and the liver is a main site of PIIINP uptake and degradation ^[18][269]. Accordingly, extramedullary clonal myeloproliferation may contribute to the increased serum concentration of PIIINP in patients with massive myeloid metaplasia of the liver.

Concerning the relationship between the serum levels of PIIINP and disease activity or disease progression and the degree of bone marrow fibrosis in PMF and related neoplasms ^{[1][6][10][11][22]}[81,86,91,261,262], it is important to realize that serum PIIINP is a marker of current organ fibrogenesis (disease activity), whereas fibrosis in the light microscope reflects the amount of previously deposited collagen (disease stage). This concept is substantiated by studies of liver patients, showing equally raised serum values in patients with fatty liver and inactive cirrhosis ^[20][271] and no relationship between serum PIIINP and the severity of liver fibrosis ^[20][271]. However, a relationship was found between serum PIIINP and ultrastructural, but not light microscopic fibrosis, which might indicate that raised serum PIIINP levels reflect early collagen formation in the liver ^[20][271].

Circulating carboxyterminal telopeptide of type I collagen (ICTP) in MPNs as a biomarker of type I collagen degradation has been found to be significantly elevated in patients with MPNs as a group compared with healthy controls ^[25][272]. However, subgroup analysis displayed only significantly elevated ICTP levels in patients with PV and MF. Of note, in this study, a significant correlation between the biomarker of collagen synthesis—PIIINP—and ICTP was recorded, reflecting that the aberrant collagen metabolism in MPNs is characterized by both enhanced collagen synthesis and degradation, the integrated signature of “a wound that never heals“ due to persistent release in the bone marrow of several growth factors (e.g., TGF-beta, FGF, and VEGF) from the malignant hyperproliferating myeloid cells in the bone marrow. Importantly, normal levels of biomarkers of type I collagen synthesis–carboxyterminal peptide of type I collagen (S-PICP) and aminoterminal propeptide of type I collagen (S-PINP), respectively, indicating that in MPNs, collagen I metabolism is disrupted in favour of enhanced collagen I catabolism. This also highlights the importance of not just measuring any type I collagen fragment, but preferentially, a specific fragment, or neo-epitope, that may provide further and reliable information on disease activity/anabolism versus collagen degradation. In contrast, collagen III formation is increased consequent to sustained repair processes in the bone marrow microenvironment as reflected in the circulation by elevated PIIINP levels ^[26][273] and in the bone marrow as increased reticulin fibrosis, being most prominent in patients with PV and MF ^[27][95]. The catabolism of type III collagen, as can be measured using fx MMP-degraded type III collagen (C3M), remains to be explored thoroughly in MPNs, but has shown promising results in various solid tumors (see Future Research Directions) ^[28][29][135,274].

As alluded to below, the trigger event behind this phenotype shift in collagen homeostasis with significant positive correlations between plasma soluble urokinase plasminogen receptor (suPAR), and PIIINP, hyaluronan and ICTP levels might be an increased production of matrix-degrading enzymes in the bone marrow ^[30][275]. Notably, biomarkers of remodelling (PIIINP, ICTP, hyaluronan) in the bone marrow (and elsewhere) correlated significantly with circulating suPAR, reflecting the activity of the urokinase-type plasminogen activator (uPA) system, which plays an important role in matrix degradation and remodelling ^[30][275]. In contrast, another family of extracellular proteolytic enzymes, the matrix metalloproteinases (MMPs), herein 2 and 9) and tissue inhibitors of metalloproteinases (TIMPs), which are also important catalyzers of matrix degradation and remodeling (please, see below) were not correlated with circulating biomarkers of collagen metabolism ^[30][275].

The fibrotic process in PMF involves the proliferation of both fibroblasts and endothelial cells ^{[27][31][32][33]}[87,95,179,255]. Endothelial proliferation is accompanied by a marked increase in basement membrane matrix components (type IV collagen and laminin), which are deposited as continuous sheets beneath the endothelial cells in the bone marrow ^[27][31][87,95]. The increase in basement membrane structures in myelofibrosis is also reflected in the circulation, where the concentration of both type IV collagen and laminin is increased in a proportion of patients with PMF and related diseases ^[5][34][85,276].

Increased serum levels of type IV-collagen-related antigens have also been found in patients with alcoholic liver disease ^[20][35][271,277], where they reflect type IV collagen metabolism, including the formation of basement membranes beneath the endothelial cells (capillarization of the sinusoids) during the development of liver cirrhosis ^[35][277]. This process appears to occur concomitantly with the deposition of collagen in the space of Disse (collagenization) ^[36][278], which is also evidenced by a positive correlation between the serum concentration of 7S collagen antigen and PIIINP ^[20][271].

Hasselbalch et al. found a similar relationship between the serum concentration of these connective tissue metabolites in PMF, which may actually reflect that the development of bone marrow fibrosis follows a pattern similar to the development of liver fibrosis, being associated with a parallel deposition of interstitial collagen type III and basement membrane material (type IV collagen) ^[5][85].

Laminin is a non-collagenous glycoprotein, which, like fibronectin, easily adheres to cell surfaces. Plasma fibronectin has been found to be inversely related to spleen size, the lowest levels being recorded in patients with huge spleens ^[37][88]. A similar relationship has been shown between serum laminin and spleen size ^[34][276], which may explain that serum laminin values were within the normal range in patients with large spleens despite evidence of active disease. However, normal serum laminin levels have also been recorded in patients with active rheumatoid arthritis, but spleen size was not evaluated in this patient group ^[38][279]. In conclusion, interstitial collagen and basement membrane metabolism in PMF are closely interrelated, which may reflect a shared stimulus for fibroblast and endothelial proliferation.

Hyaluronan (hyaluronic acid) is a highly hydrophilic glycosaminoglycan and a main component of the ECM amorphous ground substance, being particularly prevalent in soft connective tissues ^[39][40][171,280]. Elevated serum hyaluronan (HYA) levels have been found in various inflammatory connective tissue diseases, including rheumatoid arthritis ^[41][281] and scleroderma ^[42][282]. Certain liver diseases are also associated with elevated serum HYA ^[43][283]. Hyaluronan has been shown to be synthesized by fibroblasts upon various stimuli, including platelet-derived growth factor ^[42][282], which is also assumed to be involved in the pathogenesis of bone marrow fibrosis in PMF ^{[44][45][46][47]}[284,285,286,287].

Raised serum HYA concentrations have been reported in a proportion of patients with PMF ^[48][96]. Serum HYA concentrations in patients with acute disease were significantly higher than those recorded in patients with chronic disease. Furthermore, serum HYA correlated significantly with serum PIIINP and with the leucocyte count, a close co-variation being found between serum HYA, serum PIIINP and the leucocyte count with parallel changes in patients, transforming into an accelerating disease phase and in those receiving intensive chemotherapy ^[48][96].

Several explanations may be offered for the relatively modest changes in serum HYA in patients with MPNs. The metabolism of HYA may be altered in patients with splenomegaly, implying a rapid turnover of HYA. Normally, HYA is rapidly cleared from the circulation by the liver endothelial cells ^{[49][50][51][52][53]}[288,289,290,291,292], but some circulating HYA is also taken up by the spleen and other lymphoid organs ^[53][292].

Thus, the possibility exists that splenic uptake of HYA is enhanced in patients with splenomegaly, explaining the virtually normal serum HYA concentrations in patients with large spleens, despite evidence of connective tissue activation as reflected in elevated serum PIIINP concentrations ^[48][96].

The origin of excess HYA in some myelofibrosis patients remains to be clarified. Since both fibroblasts and endothelial cells have the capacity to synthesize HYA ^[54][293], raised HYA levels in the circulation may reflect the stromal cell reaction to clonal myeloproliferation. Alternatively, excess HYA may be a marker of the malignant clone itself, having the potential for HYA synthesis and production. Thus, a heterogeneous group of glycosaminoglycans has been identified in human neutrophil granules and in Auer rods of leukemic myeloblasts ^[55][294]. In addition, proteoglycan synthesis has been demonstrated in hematopoetic progenitor cells ^[56][295]. Platelet-derived growth factor (PDGF) is implicated in the pathogenesis of PMF ^{[44][45][46][47]}[284,285,286,287]. Since PDGF and other growth factors are able to stimulate HYA production ^[42][282], it is intriguing to consider the possibility that excess HYA in the circulation in PMF is caused by the intramedullary release of growth factors from the megakaryocyte cell lineage with the ensuing stimulation of HYA production from bone marrow fibroblasts and endothelial cells. Finally, impaired metabolism of HYA owing to myeloid metaplasia in the liver might contribute to elevated HYA levels in this patient group ^[43][283]. The temporal relationship between the above connective tissue components, which are similar to those observed in inflammatory connective diseases, lends support to the concept that repair processes are taking place in the bone marrow in response to clonal myeloproliferation. Thus, the co-variation of serum HYA and serum PIIINP conceivably reflects a common underlying mechanism of connective tissue activation in PMF and related diseases. However, considering the relatively small changes in serum HYA, the clinical utility of single determinations of serum HYA is limited.

As mentioned above, fibronectin is an important connective tissue constituent, which is excessively deposited in the bone marrow in PMF ^[32][57][179,180], mainly with a perivascular distribution which supports that the elevated levels in peripheral blood originate from the bone marrow compartment. Low levels of circulating fibronectin levels in patients with myelofibrosis have been associated with large spleens ^[37][88] or binding to circulating immune complexes ^[58][59][90,92]. Alternative splicing of the FN gene results in the generation of protein variants that contain the additional isoforms—extra domain A of FN (EDA) and extra domain B of FN (EDB); FN (EDA) and FN (EDB) are associated with tissue remodeling, fibroblast differentiation, inflammation, and tumor progression ^[60][296]. Highly intriguingly, EDA-FN has been shown to sustain megakaryocyte proliferation and induces a proinflammatory phenotype in bone marrow cell niches ^[61][297]. Based upon these observations circulating plasma levels of EDA-FN have been measured in patients with primary myelofibrosis in whom the highest levels of plasma EDA-FN were recorded in patients with a homozygous JAK2V617F genotype. Furthermore, increased EDA-FN levels are associated with anemia, increased high-sensitivity C-reactive protein and bone marrow fibrosis. Since elevated plasma EDA-FN at diagnosis was also found to be a predictor of large splenomegaly (over 10 cm from the left costal margin), it was concluded that plasma EDA-FN might be a new marker of disease progression and a novel target for the treatment of splenomegaly ^[62][298].

The abnormal accumulation of ECM in any organ is determined by the balance between matrix anabolic processes and the activity of matrix-degrading enzymes (e.g., MMPs, cathepsins) and their inhibitors (e.g., TIMPs and alpha-2 macroglobulin). These processes can be studied by measuring the plasma concentrations of MMP family members (e.g., MMP-1, MMP-2, MMP-3, MMP-9) and tissue inhibitors of MMPs (TIMPs) (e.g., TIMP-1). If the normally delicate balance in tissue modelling or remodelling after tissue injury of any kind is disturbed in favour of a profibrotic state, this imbalance may be reflected by increased circulating levels of anabolic matrix metabolites and/or decreased MMPs and/or increased levels of TIMPs. Studies on plasma levels of MMPs and TIMPs in MPNs ^[30][63][64][275,299,300] have shown decreased levels of several MMPs and increased TIMPs, in particular in patients with MF, in whom the TIMP-1/MMP-9 ratio was found to be significantly higher as well ^[30][275]. Furthermore, whole-blood gene expression profiling studies have shown several MMPs and TIMP3 to be deregulated in MPNs with the significant upregulation of MMP1, MMP7, MMP8, MMP9, MMP11, MMP12, MMP14 and TIMP3 ^[65][301].