Stratifying risk of infection and response to therapy in patients with myeloma: a prognostic study

Polyclonal immunoglobulins

Levels of polyclonal immunoglobulins (Igs; i.e. normal antibodies), produced from healthy plasma cells to protect against infection, are reduced in MM and this reduction, known as immunoparesis, is typically linked with bacterial respiratory tract infections, especially Streptococcus pneumoniae, which is one of the most frequent infections that occur in these patients. 1 Total levels of polyclonal Igs may provide an indication of susceptibility to infection. However, levels of antibody against a range of pertinent bacterial antigenic targets will provide more specific information regarding protection against bacterial infections, such as S. pneumoniae, but antibody levels against these bacterial targets have yet to be measured in a randomised control trial (RCT) assessing infections. In addition, anti-viral Igs against influenza and also latent viruses represent an important biomarker of immune competence, but anti-viral Ig relationships with other aspects of immune function and patient outcomes in a RCT, such as the TEAMM trial, have yet to be explored.

Neutrophil biomarkers

Neutrophils, the first cells to arrive at the site of inflammation, are the main defence against bacteria. Neutrophils kill bacteria by phagocytosis and intracellular killing, casting extracellular traps [i.e. neutrophil extracellular traps (NETs)] to entrap and kill bacteria, and degranulation (including neutrophil elastase release). Neutrophil elastase is found in azurophilic (primary) granules that release their contents into phagocytic vesicles with little discharge outside the cell except for release from disintegrating neutrophils. Lactoferrin is found in specific (secondary) granules that both release their contents into phagocytic vesicles and the extracellular medium. Matrix metalloproteinase 9 (MMP9) is found in gelatinase (tertiary) granules that more readily release their contents into the extracellular medium than the other granules. Calprotectin is found in the cytoplasm of neutrophils and makes up one-third of neutrophilic cytosolic protein. Calprotectin is released into the extracellular medium as neutrophils disintegrate, and provides anti-bacterial and anti-fungal functions. 10 Neutrophil function reduces with ageing11,12 and is implicated in the high incidence of bacterial infections in older adults. 13 At diagnosis, < 10% of MM patients have neutropenia, but all MM patients have a bacterial infection profile (i.e. organism and site) akin to patients with neutropenia. The median age of MM patients is 70 years, and poor neutrophil function, further blunted by MM and its treatment, may be a major cause of infection. Interleukin 10 (IL-10) is regarded as an anti-inflammatory cytokine that downregulates innate immune function, including those of neutrophils. 14,15 Furthermore, the level of IL-10 has been shown to be elevated in MM patients, and correlates significantly with disease progression. 16 Therefore, IL-10 may be implicated in both susceptibility to infection and active MM and be a therapeutic target. These biomarkers of immune competence have not been studied in the context of antibiotic prophylaxis and infection in myeloma. This study aimed to use these biomarkers to characterise infection risk and guide treatment stratification.

Soluble CD138

Clinical laboratories monitor MM disease activity through measurement of monoclonal Igs by densitometry on electrophoretic strips and total protein quantitation. In addition to myeloma protein changes to monitor myeloma response, levels of soluble CD138 (solCD138; also known as syndecan-1) can be measured. CD138 is a proteoglycan that is shed from myeloma cells that has effects on tumour cell growth, survival, adhesion and invasion, and on bone cell differentiation. 17,18 Elevated levels of serum solCD138 correlate with increased tumour mass and decreased patient survival. 17,19 A large cohort study demonstrated that the level of solCD138 at presentation and its reduction with treatment were powerful prognostic indicators for overall survival in MM. 20 Measurement of levels of solCD138 in the TEAMM trial at presentation and during therapy induction will enable sensitive measurement of speed and depth of response, and help establish how levofloxacin may be associated with improved response and survival.

Cytokines

Levels of interleukin 6 (IL-6) increase in the bloodstream during bacterial and viral infection. IL-6 is also a proinflammatory cytokine and a range of studies have confirmed this cytokine promotes MM cell growth, survival and drug resistance. 5,21 Classic IL-6 signalling occurs when IL-6 binds to membrane-bound proteins IL-6 receptor (IL-6R) and glycoprotein 130 (gp130). An alternative IL-6 signalling pathway termed IL-6 trans-signalling occurs when IL-6 binds to soluble IL-6R. In the IL-6 trans-signalling pathway, IL-6R is shed from the surface of host cells during MM resulting in elevated soluble IL-6R levels, followed by the formation of soluble IL-6–IL-6R complexes that bind membrane-bound gp130 expressed on tumour cells, resulting in tumour growth;22,23 the increased levels of soluble IL-6R have been associated with a poor response to chemotherapy. 24

Interleukin 8 (IL-8), a proinflammatory chemokine, is a known inducer of angiogenesis in myeloma and has been implicated in the pathophysiology of the disease. 25,26 Levels of this chemokine correlate with MM disease activity27 and IL-8 has also been shown to contribute to bone marrow-induced NF-κB (nuclear factor κ light chain enhancer of activated B cells) activity; this activity is associated with treatment resistance to bortezomib. 28 During inflammation and infection, IL-8 plays a role in the recruitment and activation of neutrophils. 29 These cytokines (i.e. IL-6 and IL-8) have been separately acknowledged in their role in infection and MM disease activity; however, their potential simultaneous roles in infection-induced inflammation, active MM and treatment resistance have yet to be investigated. Investigating these biomarkers in the TEAMM trial would uncover their importance for these interactions, and thereby help identify patients who may be at risk of poor treatment response, who may be most susceptible to infection, and those patients who may most benefit from levofloxacin and/or additional treatment pathways during induction therapy.

Procalcitonin

Procalcitonin (PCT) is a peptide precursor of the hormone calcitonin and levels become elevated in response to proinflammatory stimuli of bacterial origin, which is of key interest in newly diagnosed MM patients and for the TEAMM trial. It is considered the most sensitive biomarker of bacterial infection, correlating well with degree of sepsis, and can be used to differentiate between viral and bacterial infections in 91% of cases. 30 Levels of PCT have been shown to be able to discriminate between underlying MM disease and bacterial infection in patients where IL-6 and C-reactive protein (CRP) are linked to active MM. 31 The TEAMM trial already assesses serum CRP level as a biomarker of inflammation. A recent systematic review of RCTs has shown that measurement of PCT to confirm bacterial infection is generally associated with reduced antibiotic use, without an increase in mortality or treatment failure. 32 The use of PCT in differential diagnosis of febrile states in MM has been documented in case studies,33 but the use of PCT to confirm bacterial infection in a study as large as the TEAMM trial has not taken place.

The objective of the TEAMM trial was to tackle the early mortality rate in MM through infection prevention via antibiotic prophylaxis. This EME project built on the body of existing research by investigating the links between the triad of MM disease activity, immune competence and infection, to better understand the relationship between the intervention and patient outcomes in the TEAMM trial. Examining sensitive biomarkers of MM activity, response to therapy and bacterial infection supported this objective and provided additional benefits for patient monitoring and stratification.

Multibeads Luminex assays for anti-bacterial antibodies and cytokine measures

Serum anti-bacterial antibodies were measured using an in-house multibeads Luminex® (R&D Systems, Minneapolis, MN, USA) assay currently used in the laboratory for routine clinical assessments of immune competence and vaccination response. The assay measured immunoglobulins against 19 antigens (Table 2). Assay development, methodology and validation has been published previously. 34 For IgG the following protective thresholds were applied: 0.35 µg/ml for pneumococcal antibody serotypes,35 2 µg/ml for meningococcal antibody serotypes,36 1.0 µg/ml for H. influenza type b antibody37 and 0.1 International Unit (IU)/ml (long term) for tetanus and diphtheria antibodies. 38

Polysaccharide antigens were pre-conjugated to poly-L-lysine (PLL) and purified. PLL-conjugated antigens and toxoids were conjugated to carboxylated microspheres, each corresponding to a different bead region for the reading on a Luminex instrument. Antigen concentrations were based on previously published information and conjugated bead count was performed using a haemocytometer. The analysis of serum samples for the 19 antigens (19-plex) on multibeads was performed in a 96-well filter plate. A 12-point standard curve was obtained following a 12-fold 1 : 2 dilution of standard serum 007 in 1% phosphate-buffered saline (PBS), 0.05% bovine serum albumin (BSA) and TWEEN^® 20 (Sigma-Aldrich, St Louis, MO, USA) containing 5 µg/ml pneumococcal cell wall polysaccharide. Serum samples were diluted 1 : 100 using 1% PBS, 0.05% BSA, and TWEEN 20 containing 5 µg/ml pneumococcal cell wall polysaccharide and 5 µg/ml pneumococcal serotype 22F to prevent unspecific binding. Human serum, pooled serum and serum deprived of IgGs were used as controls.

For cytokine analysis a commercially available kit was purchased from Bio-Rad Laboratories, Inc. (Hercules, CA, USA) and assays were performed following the manufacturer’s instructions. Pre-made bead-conjugated anti-IL-6, -IL-8 and -IL-10 antibodies and standards for each cytokine were included in the kit. Pooled sera from eight healthy donors were used as controls.

Beads were pre-adsorbed on the filter plate and then incubated (at room temperature) in the dark on an orbital shaker [at 500 revolutions per minute (r.p.m.)] with standards, controls or samples. Anti-human IgG antibody and a reporting system (i.e. phycoerythrin) were then added to the samples and further incubated (at room temperature) in the dark on an orbital shaker (at 500 r.p.m.). Reading and data analysis were performed on a Luminex instrument running the Bio-Plex Manager™ software (version 6.1; Bio-Rad Laboratories, Inc., Hercules, CA, USA). Coefficients of inter-plate variation are shown in Table 3.

Enzyme-linked immunosorbent assays for the detection of neutrophil biomarkers, CD138 and procalcitonin

Detection and measurement of serum levels of neutrophil biomarkers, solCD138 and procalcitonin in the TEAMM trial patients were performed by sandwich enzyme-linked immunosorbent assays (ELISAs). Kits for the analysis of neutrophil elastase, MMP9, lactoferrin, solCD138 and procalcitonin were purchased from Abcam (Cambridge, UK). Assays were performed following manufacturers’ instructions and guidance.

A modified capture ELISA for the quantification of NETs was performed following the protocol published by Caudriller et al. 39 The capture ELISA is based on the association of myeloperoxidase (MPO) with the deoxyribonucleic acid (DNA) of NETs. Briefly, the anti-MPO capture antibody was coated on a 96-well plate overnight at 4 °C and, after washing, serum samples were added. A peroxidase-labelled anti-DNA antibody was added as a secondary antibody; the peroxidase substrate was included as the reporting system.

Calculated coefficients of inter- and intra-plate variation for all assays are shown in Table 3.

Haemagglutination assay for anti-viral antibody measures

Anti-viral antibodies were measured in the serial samples of 19 TEAMM trial patients who received flu vaccines during the trial period. Antibodies against influenza B, influenza A strains H1N1 and H3N2 were measured before and after the TEAMM trial patients had received their flu vaccine.

Haemagglutination assays (HAIs) were performed by the Division of Virology at the National Institute for Biological Standards and Control (NIBSC; Potters Bar, UK). The units of measurement of anti-viral antibodies are HAI titres, in which a HAI titre of ≥ 40 is correlated with protection against influenza infection. 40

Statistical analysis

The main analysis, to determine any associations between outcome measures and biomarkers using prespecified cut-off points were explored using, where appropriate, chi-squared tests for categorical variables using Fisher’s exact test. Mann–Whitney U-test, Friedman and Dunn’s modified test were used to compare categorical data between groups. Correlations between biomarkers and outcomes measures were investigated using Spearman’s rank test.

The analysis of time to first infection within 12 weeks was carried out using a log-rank comparison, starting from the date the patient started trial treatment to the date of first infection, or to a censor date for those patients with no infections. All randomised patients were included in an intention-to-treat analysis and assessed using Kaplan–Meier curves and the log-rank test. 41 Overall survival was calculated from date of starting trial treatment to the date of death or censor, as appropriate. Cox proportional hazards models were utilised to compare factors after adjustment for age and treatment pathway. 42 Statistical analyses were performed using IBM^® SPSS Statistics (version 21; IBM, Portsmouth, UK), Prism (version 8; GraphPad, GraphPad Software Inc., San Diego, CA, USA) and SAS^® (version 9.3; SAS Institute Inc., Cary, NC, USA) software.

Total levels of polyclonal IgG, IgA and IgM

Suppression of polyclonal immunoglobulin levels (immunoparesis) is a hallmark of myeloma. More than 80% of myeloma patients are expected to have levels of polyclonal immunoglobulins below the normal range,43 which contributes to the increased susceptibility to infections identified in patients. Levels of polyclonal IgG, IgA and IgM were measured in TEAMM trial participants without an IgG, IgA and IgM myeloma monoclonal protein (m-protein), respectively:

• polyclonal IgG levels –

  • in healthy controls: 10.17 g/l

  • TEAMM trial participants at baseline: 4.40 g/l

  • TEAMM trial participants at 12 weeks: 3.44 g/l

  • TEAMM trial participants at 1 year: 6.65 g/l

• polyclonal IgA levels –

  • in healthy controls: 2.28 g/l

  • TEAMM trial participants at baseline: 0.32 g/l

  • TEAMM trial participants at 12 weeks: 0.28 g/l

  • TEAMM trial participants at 1 year: 0.55 g/l

• polyclonal IgM levels –

  • in healthy controls: 0.90 g/l

  • TEAMM trial participants at baseline: 0.17 g/l

  • TEAMM trial participants at 12 weeks: 0.18 g/l

  • TEAMM trial participants at 1 year: 0.28 g/l.

At baseline, the majority of TEAMM trial patients had levels of polyclonal IgG (76%), IgA (83%) and IgM (90%) below the normal range (Figure 4). Median levels of polyclonal IgG were further decreased after 12 weeks of therapy, whereas polyclonal IgA and IgM levels remained unaltered compared with baseline. After 1 year, polyclonal levels for all three types of immunoglobulins had increased compared with baseline. The median levels of polyclonal IgG were within the normal range, recovering above levels at diagnosis. However, despite levels increasing above baseline, median levels of IgA and IgM remained below the normal range at 1 year. For all three types of polyclonal immunoglobulins median levels in the TEAMM trial patients at all time points were significantly lower than median levels identified in the healthy cohort.

FIGURE 4.

Levels of polyclonal immunoglobulins in the TEAMM trial and the healthy controls. (a) Polyclonal IgG levels; (b) polyclonal IgA levels; and (c) polyclonal IgM levels. The distribution and median levels (red lines) of polyclonal immunoglobulins IgG, IgA and IgM in a healthy cohort and in TEAMM trial patients at baseline, after 12 weeks of therapy and at 1 year. For TEAMM trial patients, IgG levels are from those patients who do not have an IgG m-protein; IgA levels are from those patients who do not have an IgA m-protein; and IgM levels are from those patients who do not have an IgM m-protein. The fifth and 95th percentile of normal ranges identified in healthy conditions are shown as light blue dotted lines. Statistical significant differences between TEAMM trial patients’ median values and healthy controls’ median values is represented with # (i.e. a p-value of < 0.001 = ###), whereas statistical significance between TEAMM trial time points are indicated with * (not significant = ns; a p-value of < 0.001 = ***).

Serum IgG antibody levels against bacterial antigens commonly associated with infections (i.e. pneumococcus, meningococcus, Haemophilus influenza type b, tetanus and diphtheria toxoid)

Anti-bacterial antibody levels against commonly encountered bacterial antigens have been quantified at baseline for 841 of the TEAMM trial patients. This analysis is the largest of its kind in a cohort of myeloma patients and provided novel information regarding immune competence in myeloma and the severity of impaired antibody defence that patients have against bacterial infections. The World Health Organization (WHO) provides guidelines regarding the levels of anti-bacterial antibodies expected in individuals for protection against infections. The distribution of anti-bacterial antibodies in the TEAMM trial is shown in Figure 5 along with the WHO’s protective threshold level as discontinuous lines. The median antibody levels of TEAMM trial patients were below the recommended protection levels for 18 of the 19 bacterial antigens tested. The only exception was identified for the bacterial pneumococcal serotype 14, against which the median level of antibodies in TEAMM trial patients was comparable to the WHO’s protective level; however, this still left 40% of patients without protection.

FIGURE 5.

Levels of anti-bacterial antibodies in TEAMM trial patients at disease presentation compared with thresholds of protection against bacterial infections. A scatterplot showing the distribution of anti-bacterial immunoglobulins in 841 TEAMM trial patients at disease presentation and their cumulative median level (in red) along with the WHO’s protective threshold level as discontinuous lines. At baseline, myeloma patients are immunocompromised and present levels of anti-bacterial immunoglobulins below the WHO-recommended threshold. (a) Pn1 IgG; (b) Pn3 IgG; (c) Pn4 IgG; (d) Pn5 IgG; (e) Pn6B IgG; (f) Pn7F IgG; (g) Pn9V IgG; (h) Pn14 IgG; (i) Pn18C IgG; (j) Pn19A IgG; (k) Pn19F IgG; (l) Pn23F IgG; (m) HiB IgG; (n) tetanus IgG; (o) diphtheria IgG; (p) Men A IgG; (q) Men C IgG; (r) Men W-135 IgG; and (s) Men Y IgG. HiB, H. influenza type b; men, meningococcal; Pn, pneumococcal.

Antibody protection was previously characterised against these infective agents in a cohort of 193 healthy volunteers. 34 Compared with this healthy cohort (Table 4), a significantly lower percentage of TEAMM trial patients demonstrated protective levels for all bacterial antigens. Typically, > 55% of healthy donors exhibited protective levels against any of the bacterial targets compared with < 20% of myeloma patients. Pneumococcal (Pn) serotypes Pn1, Pn3 and Pn4 had strikingly low protection rates with < 10% of myeloma patients having the minimum level required for protection.

In clinical assessment of response to pneumococcal vaccines containing multiple serotypes (typically 13–23) in healthy individuals a response is considered normal when inducing protective levels against two-thirds of serotypes (8/12). The healthy controls in this study were UK blood donors under the age of 65 years and thus unlikely to have ever been vaccinated against pneumococcus. Therefore, these blood donors’ antibody levels against pneumococcal serotypes are the result of natural exposure. By contrast, TEAMM trial patients over the age of 65 years were likely to have received the 23-valent polysaccharide pneumococcal vaccine PPV23 (PNEUMOVAX^® 23; Merck, Sharp & Dohme, Kenilworth, NJ, USA) as part of the UK NHS vaccine schedule administered in primary care to all individuals in the autumn in their 66th year. Notably, only 6% of myeloma patients had protective levels for two-thirds of pneumococcal serotypes compared with 55% of healthy blood donors. Indeed, only 15% and 33% of patients had protective levels for half and one-third of serotypes, respectively. The highest level of protection for the healthy donors was found for tetanus, in which 95% of individuals displayed protective levels in contrast to 41% of myeloma patients. Even for the tetanus toxoid, the levels of toxoid-specific IgG were lower than total IgG levels (i.e. 4% vs. 43% of normal median levels). For 18 out of 19 bacterial antigen targets, levels of specific IgG were substantially lower than the levels of total polyclonal IgG.

Adaptive immunity wanes as age increases above 50 years and in people over 65 years of age antibody response to vaccination is substantially less than in younger individuals. Antibody protection against bacterial antigens was found to differ between patients < 65 and patients ≥ 65 years of age (Table 5). As expected, proportionately more young patients had protective levels against meningococcal serotypes, H. influenza type b, and tetanus and diphtheria toxoids. However, a higher proportion of older individuals showed protective levels against pneumococcal serotypes than of younger patients. These differences between antigens based on age reflect the UK’s PPV23 vaccination programme for individuals aged ≥ 65 years. Although these findings provide some support for the efficacy of the current vaccination programme in this age/patient population, significant differences were found for only three pneumococcal serotypes. Furthermore, the majority of older patients still failed to meet protective levels and only 6% of patients demonstrated protective levels for 8 out of 12 serotypes, which is a normal adult response to PPV23 vaccination.

Correlation of functional antibody levels with infections, febrile episodes and deaths

The association among antibody levels, prevalence of infections, febrile episodes and deaths in the TEAMM trial was investigated. The TEAMM patients were divided into tertiles or quartiles based on their levels of anti-bacterial antibodies to explore if those patients with the lowest levels had a higher incidence of febrile episodes and/or were more likely to die within the year following diagnosis. Frequencies of febrile episodes and deaths in these groups of patients are shown in Figure 6. Levels of anti-pneumococcal serotype 19A < 0.06µg/ml in TEAMM trial patients were associated with an increased incidence of febrile episodes and deaths (25% and 13%, respectively) up to 12 weeks after the beginning of treatment, compared with patients with levels above 0.06 µg/ml (16% and 12%, respectively) and with patients with levels above the WHO-recommended threshold of 0.35 µg/ml (20% and 11%, respectively). Other associations were found between low levels of antibodies against pneumococcal serotypes and occurrence of deaths. Levels of anti-pneumococcal serotype 5 < 0.1 µg/ml were associated with an increased incidence of deaths, > 37% compared with < 15% in patients with levels ≥ 0.1 µg/ml. Similarly, deaths were more frequent in patients with levels of anti-pneumococcal serotype 7F < 0.13 µg/ml (i.e. in 17% of patients) than in patients with levels ≥ 0.13 µg/ml (i.e. in 10% of patients). Low levels of antibodies against other bacterial antigens not part of the pneumococcal family were also found to be associated with an increased numbers of deaths in the TEAMM trial at 1 year. Deaths were more frequent in patients with antibody levels against H. influenza type b < 0.2 µg/ml (i.e. 39% of patients) than in patients with levels of antibodies above this threshold (≤ 23%). Similarly, lower levels of antibodies against meningococcal serotype A were found to be associated with the occurrence of death (53% vs. ≤ 34%) in the TEAMM trial.

FIGURE 6.

Frequencies of febrile episodes and deaths in TEAMM trial patients stratified into groups based on their levels of anti-bacterial antibodies. Frequencies of (a) febrile episodes and (b) deaths in TEAMM trial patients with levels of anti-Pn19A antibodies < 0.06 µg/ml, between 0.06 µg/ml and 0.35 µg/ml, and ≥ 0.35 µg/ml; (c) frequency of deaths in TEAMM trial patients with levels of anti-Pn7F antibodies < 0.13 g/ml, between 0.13 g/ml and 0.35 µg/ml, and ≥ 0.35 µg/ml; (d) frequency of deaths in TEAMM trial patients with levels of anti-Pn5 antibodies < 0.03 µg/ml, between 0.03 µg/ml and 0.1 µg/ml, between 0.1 µg/ml and 0.35 µg/ml, and ≥ 0.35 µg/ml; (e) frequency of deaths in TEAMM trial patients with levels of anti-Men A antibodies < 0.5 µg/ml, between 0.5 µg/ml and 2 µg/ml, and ≥ 2 µg/ml; and (f) frequency of deaths in TEAMM trial patients with levels of anti-HiB antibodies < 0.2 µg/ml, between 0.2 µg/ml and 0.5 µg/ml, between 0.5 µg/ml and 1 µg/ml, and ≥ 1µg/ml. Statistical significance was calculated with Fisher’s exact test (*p < 0.05; **p < 0.01; ***p < 0.001). HiB, H. influenza type b; men, meningococcal.

Anti-viral antibodies

Levels of anti-viral antibodies were measured in the samples of a sub-cohort of 19 of TEAMM trial patients who received flu vaccines during the trial period. Antibodies against influenza B, influenza A strain H1N1 and influenza A strain H3N2 were measured in 19 patients before and after they received flu vaccine and, again, at 1 year (range 9–15 months) after vaccination. A threshold HAI titre ≥ 40 was used to indicate protection against flu strains (i.e. discontinuous line in Figure 7). Before vaccination, the median antibody level for the 19 TEAMM trial patients was below the recommended threshold for protection. In fact, the percentages of patients protected against these viral strains were 0% for influenza B, 5% for H1N1 and 32% for H3N2. After receiving vaccination, antibody titres increased in a proportion of patients, the increase was statistically significant for strains H1N1 and H3N2, and the percentage protected increased to 26%, 42% and 58% for flu strains influenza B, H1N1 and H3N2 (see Figure 7 and Table 6), respectively. The study was also able to measure anti-viral antibodies in 14 of the initial 19 patients at 1 year after beginning the trial and receiving flu vaccination. The median level of antibody titre had significantly decreased by 1 year. The percentage of protection against all three viral strains decreased to 14%, indicating that only two patients for each strain were protected against flu virus infections. Interestingly, the two patients showed protection against both strains H1N1 and H3N2 at 1 year, although the patients showed a HAI titre of anti-influenza B antibodies equal to zero.

FIGURE 7.

Distribution of anti-viral antibodies in vaccinated TEAMM trial patients before and after vaccination and at 1 year after the first trial visit. Levels of antibodies against influenza B, H1N1 and H3N2 strains in 19 TEAMM trial patients who received flu vaccination during the trial. The number of patients with undetected levels of antibodies (i.e. with a HAI titre = 0) is shown in red and the HAI titre threshold recommended for protection against viral strains (i.e. a HAI titre = 40) is represented as a discontinuous line. (a) Distribution, average and standard deviation of levels of antibodies before vaccination, after vaccination and at 1 year post disease presentation for influenza B; (b) distribution, average and standard deviation of levels of antibodies before vaccination, after vaccination and at 1 year post disease presentation for H1N1; (c) distribution, average and standard deviation of levels of antibodies before vaccination, after vaccination and at 1 year post disease presentation for H3N2; (d) before-and-after vaccination graphs representing levels of antibodies for each vaccinated TEAMM trial patient for influenza B; (e) before-and-after vaccination graphs representing levels of antibodies for each vaccinated TEAMM trial patient for H1N1; and (f) before-and-after vaccination graphs representing levels of antibodies for each vaccinated TEAMM trial patient for H3N2. Statistical significance was calculated with Dunn’s test (*p < 0.05; ***p < 0.001). ns, not significant.

In order to identify possible correlations between viral and bacterial protection, anti-flu antibodies were analysed in association with anti-bacterial antibodies data identified in the TEAMM trial patients post vaccination. Correlation values and statistical significance for each pair of antigens are summarised in Table 7. A strong correlation was identified between levels of antibody against influenza type b and Pn19A (highlighted in purple in Table 7). Moderate associations (highlighted in light blue in Table 7) were also identified between anti-viral influenza B and anti-Pn18C and -Pn19F; and between anti-H1N1 and anti-Pn18C, -Pn19F, -influenza B and diphtheria. Based on these data, antibody protection against bacterial pathogens was evaluated on flu-vaccinated TEAMM trial patients before and after their vaccination (Table 8). Although for most of the bacterial serotypes the level of protection remained unchanged or lower, protection against two serotypes increased in flu-vaccinated TEAMM trial patients, and these are Pn4 and tetanus (as highlighted in purple in Table 8).

Cytokine interleukin 10 levels

The cytokine IL-10 is known to be elevated in myeloma and is associated with worse survival. 44,45 Levels of IL-10 were measured in the sera in a cohort of 16 healthy individuals and in 321 TEAMM trial patients at diagnosis. The concentration of IL-10 in the serum of TEAMM trial patients has been found to exceed the level identified in healthy individuals, which was below the limit of detection of the assay (Figure 9). In fact, at baseline, 95% of sera analysed of TEAMM trial patients had IL-10 levels above the level identified in healthy controls (1.74 pg/ml). Analysis was performed in order to verify whether high levels of IL-10 in myeloma patients was associated with a higher prevalence of non-febrile infections and/or febrile episodes. The results summarised in Table 10 show that one-third of the cohort of patients with high levels of IL-10 (i.e. ≥ 10 pg/ml) had a greater frequency of infections (57%) and febrile episodes (25%) than the proportion of TEAMM trial patients with lower levels of IL-10 (i.e. < 10 pg/ml) who showed a lower incidence of episodes [i.e. < 40% and 21%, respectively, for infection episodes (febrile episodes and/or non-febrile episodes) and febrile episodes only]. The association between levels of IL-10 ≥ 10 pg/ml and incidence of febrile episode, in respect to patients with a IL-10 level of < 7, ≥ 7 and < 10 pg/ml, was also confirmed by Fisher’s exact test. The results (Figure 10) showed statistical significance at a p-value < 0.0001. The association between IL-10 levels in TEAMM trial patients at baseline and response to therapy, in terms of percentage change of free light chains (FLCs) between trial visits, was evaluated. No correlation was found between IL-10 levels and response to therapy at any trial visits up to 1 year (rho < 0.25).

FIGURE 9.

Baseline levels of cytokine IL-10 in TEAMM trial patients compared with levels identified in the healthy cohort. Box-and-whisker plots representing medians, 25–75th percentiles, and minimum and maximum values of IL-10 levels in 16 healthy volunteers and in 321 TEAMM trial patients. Statistical significance was calculated with the Mann–Whitney U-test (p-value < 0.001 = ***).

FIGURE 10.

Frequencies of febrile episodes in TEAMM trial patients divided in to groups based on their levels of IL-10. Frequencies of febrile episodes in TEAMM trial patients with IL-10 levels < 7 pg/ml, between 7 pg/ml and 10 pg/ml, and ≥ 10 pg/ml. Statistical significance was calculated with Fisher’s exact test (p-value < 0.001 = ***).

Analysis of risk of infection and survival in TEAMM trial patients based on immune competence biomarkers at baseline

Based on the analysis previously shown (see Correlation of functional antibody levels with infections, febrile episodes and deaths and Cytokine interleukin 10 levels), biomarkers anti-Pn5, -Pn7F, -Pn19A, -meningococcal (Men) A antibodies and IL-10 along with polyclonal immunoglobulins were selected as possible predictors of febrile/non-febrile episodes and survival among TEAMM trial patients. TEAMM trial patients were divided into tertiles or quartiles based on their levels of biomarker to explore if those patients with the lowest levels had a higher risk of febrile/non-febrile episodes and/or if they were more likely to die within the year following diagnosis.

Levels of soluble CD138

The analysis of serum levels of solCD138 was performed for a cohort of 327 TEAMM trial patients at all trial time points and compared with levels measured in 50 healthy donors (age range 17–65 years). As expected, median levels of solCD138 were elevated in TEAMM trial patients, compared with the healthy cohort (TEAMM trial: 120.4 IU/ml, range 1.41–256 IU/ml; healthy cohort: 36.6 IU/ml, range 18.49–57.42 IU/ml), confirming its validity as a biomarker of active disease (Figure 11). In total, 28% of TEAMM trial patients presented with solCD138 levels within the normal range, whereas 72% of patients had a solCD138 level higher than the highest value in the 50 healthy controls (Table 19). At diagnosis, in the 234 patients with solCD138 levels above the normal range, parallel to m-protein and FLC levels, the solCD138 levels decreased thereafter in association with the anti-myeloma induction therapy. However, average levels of solCD138 in TEAMM trial patients remained elevated compared with the normal range at all trial time points to 1 year of follow-up (Figure 12). During induction therapy, levels of solCD138 decreased by 25% during the first 4 weeks of treatment, 25% between 4 and 8 weeks, 12% between 8 and 12 weeks, 7% between 12 and 16 weeks and 17% between 16 and 52 weeks (Table 20). By 1 year, most patients were in remission.

FIGURE 11.

Levels of solCD138 in TEAMM trial patients at presentation compared with levels in a healthy cohort. Box-and-whisker plots representing medians, 25–75th percentiles, and minimum and maximum values of levels of solCD138 in 45 healthy volunteers and 327 TEAMM trial patients. Statistical significance was calculated with the Mann–Whitney U-test (p-value < 0.001 = ***).

FIGURE 12.

Levels of solCD138 in TEAMM trial patients at each trial visit up to 1 year from presentation. Average levels and standard deviations of solCD138 in TEAMM trial patients at disease presentation, 4 weeks, 8 weeks, 12 weeks, 16 weeks and 1 year after presentation. Levels of solCD138 identified in 45 healthy controls are the normal range shown in green: median level (solid line), 5th and 95th percentile (discontinued lines).

Procalcitonin

Procalcitonin levels were measured in samples of a proportion of TEAMM trial patients collected at all four time points of the TEAMM trial. The study’s results indicated an average of 6.4% of investigated TEAMM trial patients presented elevated PCT levels at each time point, whereas the remaining patients had undetectable levels. The study selected patients who had samples collected 1 day before or up to 3 days after the occurrence of a febrile or non-febrile episode. Levels of PCT were analysed in a total of 97 patients, of whom 33% had at least one febrile episode during the trial and 67% had other types of non-febrile episodes. The concentration used as threshold for inflammation/infection was a PCT level ≥ 0.15 ng/ml. The study’s results indicate that PCT levels were elevated in only 50% of the cases for both febrile and other infection episodes.

C-reactive protein

Levels of CRP were also measured in all TEAMM trial patients, at all time points, as a biomarker of inflammation and bacterial infection. Levels of CRP were above the threshold for inflammation/infection (i.e. ≥ 10 mg/l) in 60% of patients, specifically CRP levels were elevated in 60% of patients who had febrile episodes and 57% of patients who had non-febrile episodes. Analysis of PCT levels in association with CRP levels showed no correlation between the two parameters.

Cytokines interleukins 6 and 8

The analysis of serum proinflammatory cytokines IL-6 and IL-8, as biomarkers of infection-induced inflammation, was performed for ≈300 TEAMM trial patients and compared with levels identified in the sera of a cohort of 16 healthy controls, in which blood samples were separated 24 hours after venesection. Surprisingly, the analysis indicated that levels of IL-6 and IL-8 in TEAMM trial patients at presentation were below those measured in the healthy cohort and the concentration measured at 12 weeks and at 1 year post treatment showed no statistically significant difference compared with levels identified before treatment (Figure 15). Although a moderate association was found between levels of IL-6 and IL-8 at all the study time points (rho = 0.6), no correlation was identified for these cytokines with the biomarker of infection, that is, CRP, at baseline (rho < 0.25). Furthermore, data analysis was performed in order to verify whether the distribution of IL-6 and IL-8 at baseline was different in TEAMM trial patients who had febrile and/or non-febrile episodes or none during the trial (Figure 16). Results indicated that levels of IL-6 and IL-8 were comparable between patients having febrile and/or non-febrile episodes and patients who had no episodes during the trial. Statistically significant differences were identified in only IL-6 levels between patients who had non-febrile episodes only and patients who had fevers and non-febrile episodes during the trial, suggesting that patients more prone to either type of episodes might have lower levels of IL-6. However, no statistically significant difference was observed compared with patients who had no episodes during the trial.

FIGURE 15.

Median levels of (a) IL-6 and (b) IL-8 in TEAMM patients at disease presentation, 12 weeks and 1 year after presentation. Levels of IL-6 and IL-8 in 16 healthy controls are shown as normal range in light blue: median level, maximum and minimal values (discontinued lines). Statistical significance was calculated with Dunn’s unmodified test for multiple comparison (a p-value of < 0.05 = *; not significant = ns).

FIGURE 16.

Levels of cytokines (a) IL-6 and (b) IL-8 at presentation in TEAMM trial patients with different infection profiles during the trial. Box-and-whisker plots representing median, 25th–75th percentiles, and minimum and maximum values of IL-6 and IL-8 concentrations in TEAMM trial patients with different infection profiles. Statistical significance was calculated with Dunn’s unmodified test for multiple comparison (a p-value of < 0.05 = *; not significant = ns).

The levels of IL-6 and IL-8 before and after therapy, and at 1 year in TEAMM trial patients was also compared between good myeloma responders and poor responders (Figure 17). The TEAMM trial patients were categorised as good responders if showing ≥ 90% reduction in FLCs at 16 weeks, whereas poor responders were patients with < 90% reduction in FLCs at 16 weeks. Both groups of patients had levels of IL-6 and IL-8 below the normal range at all trial visits. Interestingly, patients achieving a poor response to therapy at 16 weeks also showed lower levels of IL-6 at baseline than good responders; however, this difference was lost by 12 weeks and at 1 year. Statistically significant differences were not found between levels of IL-8 in the two response groups; however, the median level in poor responders at 12 weeks was notably lower than baseline levels in the same group of response and also compared with levels post therapy in good responders.

FIGURE 17.

Distribution and median levels of cytokines (a) IL-6 and (b) IL-8 at baseline, 12 weeks after therapy and at 1 year in TEAMM trial patients divided by their response to therapy. In dark blue good responders and in grey poor responders are shown. Levels of IL-6 and IL-8 in 16 healthy controls are shown as normal range in light blue: median level, maximum and minimal values (discontinued lines). Statistical significance between trial time points was calculated with Dunn’s unmodified test for multiple comparison (no significance found), whereas statistical significance between response groups at each time point was calculated with the Mann–Whitney U-test (a p-value of < 0.05 = *; not significant = ns).