Global survival trends for brain tumors, by histology: analysis of individual records for 556,237 adults diagnosed in 59 countries during 2000-2014 (CONCORD-3)

BACKGROUND

Survival is a key metric of the effectiveness of a health system in managing cancer. We set out to provide a comprehensive examination of worldwide variation and trends in survival from brain tumors in adults, by histology.

METHODS

We analyzed individual data for adults (15-99 years) diagnosed with a brain tumor (ICD-O-3 topography code C71) during 2000-2014, regardless of tumor behavior. Data underwent a 3-phase quality control as part of CONCORD-3. We estimated net survival for 11 histology groups, using the unbiased nonparametric Pohar Perme estimator.

RESULTS

The study included 556,237 adults. In 2010-2014, the global range in age-standardized 5-year net survival for the most common sub-types was broad: in the range 20%-38% for diffuse and anaplastic astrocytoma, from 4% to 17% for glioblastoma, and between 32% and 69% for oligodendroglioma. For patients with glioblastoma, the largest gains in survival occurred between 2000-2004 and 2005-2009. These improvements were more noticeable among adults diagnosed aged 40-70 years than among younger adults.

CONCLUSIONS

To the best of our knowledge, this study provides the largest account to date of global trends in population-based survival for brain tumors by histology in adults. We have highlighted remarkable gains in 5-year survival from glioblastoma since 2005, providing large-scale empirical evidence on the uptake of chemoradiation at population level. Worldwide, survival improvements have been extensive, but some countries still lag behind. Our findings may help clinicians involved in national and international tumor pathway boards to promote initiatives aimed at more extensive implementation of clinical guidelines.

Formal Error Rate in a Computerized Obstetric Medical Record

Missing and incompatible data rates in a computerized obstetric medical record are described and range from 0-84% with a median of 6%. Comparison is made to previous manual records and missing data rates appear to be similar. Certain unique characteristics of each variable checked (logistical problems, data sheet format, and low frequency, expected positive response) seem to have a stronger influence on the missing rate than do categories of time collected, source, type and transmission mode. There is bias in missing rates according to different patient types, but incompatible data rates appear not to be influenced.

Computerized Nursery Discharge Summary

Expansion of a computerized obstetric medical record to include newborn nursery information is described. Data sheets are entered for each infant discharged, and a computer-typed discharge summary is generated which abstracts information originally gathered by both obstetric and pediatric medical personnel. The system has improved communication with medical providers who administer follow-up care after discharge and has replaced some physician dictation effort, especially in the Full Term Nursery. In the Intensive Care Nursery the summary has not been successful because data collected is not felt adequate to describe the chronology of a long hospitalization. Comparison of the computer record and manual discharge diagnosis sheet is presented.

Acute G-CSF therapy is not protective during lethal E. coli sepsis

We investigated whether decreases in circulating polymorphonuclear neutrophils (PMN) during lethal Escherichia coli (E. coli) sepsis in canines are related to insufficient host granulocyte colony-stimulating factor (G-CSF). Two-year-old purpose-bred beagles had intraperitoneal E. coli-infected or -noninfected fibrin clots surgically placed. By 10 to 12 h following clot, both infected survivors and nonsurvivors had marked increases (P = 0.001) in serum G-CSF levels (mean peak G-CSF ng/ml +/- SE, 1,931 +/- 364 and 2,779 +/- 681, respectively) compared with noninfected controls (134 +/- 79), which decreased at 24 to 48 h. Despite increases in G-CSF, infected clot placement caused delayed (P = 0.06) increases in PMN (mean +/- SE change from baseline in cells x 10(3)/mm(3) at 24 and 48 h) in survivors (+3.9 +/- 3.9 and +13.8 +/- 3.6) compared with noninfected controls (+13.1 +/- 2.8 and +9.1 +/- 2.5). Furthermore, infected nonsurvivors had decreases in PMN (-1.4 +/- 1.0 and -1.1 +/- 2.3, P = 0.006 compared with the other groups). We next investigated whether administration of G-CSF immediately after clot placement and continued for 96 h to produce more rapid and prolonged high levels of G-CSF after infection would alter PMN levels. Although G-CSF caused large increases in PMN compared with control protein from 2 to 48 h following clot in noninfected controls, it caused much smaller increases in infected survivors and decreases in infected nonsurvivors (P = 0.03 for the ordered effect of G-CSF comparing the three groups). Thus insufficient host G-CSF is unlikely the cause of decreased circulating PMN in this canine model of sepsis. Other factors associated with sepsis either alone or in combination with G-CSF itself may reduce increases or cause decreases in circulating PMN.

Building knowledge in a complex preterm birth problem domain

Data mining methods used a racially diverse sample (n = 19,970) of pregnant women and 1,622 variables that were collected in Duke's TMR electronic patient record over a 10-year period. Different statistical and data mining methods were similar when compared using receiver operating characteristic (ROC) curves. Best results found that seven demographic variables yielded .72 and addition of hundreds of other clinical variables added only .03 to the area under the curve (AUC). Similar results across methods suggest that results were data-driven and not method-dependent, and that demographic variables may offer a small set of parsimonious variables with predictive accuracy in a racially diverse population. Work to determine relevant variables for improved predictive accuracy is ongoing.

Long-term safety of treatment with recombinant human granulocyte colony-stimulating factor (r-metHuG-CSF) in patients with severe congenital neutropenias

Congenital neutropenias include a heterogenous group of diseases characterized by a decrease in circulating neutrophils. In phase I/II/III studies in patients with severe congenital and cyclic neutropenia, treatment with recombinant human granulocyte colony-stimulating factor (r-metHuG-CSF) resulted in a rise in the absolute neutrophil counts (ANC) and a reduction in infections. We report the effects of long-term safety of subcutaneous r-metHuG-CSF administration in 54 patients (congenital n = 44. cyclic n = 10) treated for 4-6 years. A sustained ANC response was seen in 40/44 severe congenital neutropenia patients and 10/10 cyclic neutropenia patients. Two patients required an increase of > 25% in dose to maintain a clinical response; one patient became refractory to therapy. A significant decrease in the incidence of severe infections and the need for intravenous antibiotics was noted. Significant adverse events noted which may or may not be related to therapy included: osteopenia (n = 15), splenomegaly (n = 12), hypersplenism (n = 1), vasculitis (n = 2), glomerulonephritis (n = 1), BM fibrosis (n = 2), MDS/leukaemia (n = 3), and transient inverted chromosome 5q with excess blasts (n = 1). R-metHuG-CSF has been well tolerated in the majority of patients and resulted in a long-term improvement in their clinical status.

Mechanisms of recovery from mechanical injury of renal tubular epithelial cells

The mechanism(s) whereby a denuded renal tubular epithelial cell surface becomes reestablished remains unknown. We therefore measured the rate of renewal of mechanical wounds made in confluent monolayers of two established renal tubular epithelial cell lines. We found that wounds of MDCK cells heal at a faster rate than wounds of LLC-PK1 cells. The magnitude of wound healing did not differ when cells grown on plastic were compared with cells grown on fibronectin, laminin, or collagen. Irradiation (4,000 rads) of MDCK and LLC-PK1 cells significantly reduced indexes of proliferation (5-bromo-2'-deoxyuridine and thymidine uptake) without affecting wound healing. Serum and epidermal growth factor (EGF) enhance whereas transforming growth factor-beta 1 (TGF-beta 1) impairs wound healing. Hepatocyte growth factor (HGF) stimulates wound healing at low concentrations and inhibits healing at high concentrations in MDCK cells while not affecting healing of LLC-PK1 cell wounds at any concentration. Several interleukins (IL-1, IL-2, IL-3, and IL-6) did not affect wound healing in either cell type. Healing of LLC-PK1 but not MDCK cells was impaired by exposure to a peptide containing a RGD sequence. Conversely, healing of MDCK but not LLC-PK1 cells was impaired by the REDV tetrapeptide. Healing of both LLC-PK1 and MDCK was impaired by heparin but not by the LDVPS peptide. These results demonstrate that mechanical wounds of LLC-PK1 and MDCK cells heal, at least in part, by migration. Healing is regulated by serum and growth factors including EGF, HGF, and TGF-beta 1.(ABSTRACT TRUNCATED AT 250 WORDS)

Automated acuity scoring within a computer based medical record

This paper describes the initial development of a completely automated acuity scoring system that resides within the TMR bedside computing system at the Duke University Medical Center, Surgical Intensive Unit. The scoring system is based upon the APACHE II acuity scoring system and provides for the recalculation of acuity scoring at 12 hour intervals through the patient's ICU course. When comparing hand calculated versus computer generated acuity scores for 19 patients, discrepancies fell into three broad categories: 1) data available to the application differed from that available to the human scorer. 2) apparent transcription errors 3) data items lost or absent from the paper record. It remains to be determined if computer generated acuity scoring provides for a more accurate representation of the patient's acuity.

A new low-frequency antigen, Hga (Hughes)

This paper describes a 'new' low-frequency antigen, Hga, which came to light during routine antibody identification tests. 3 families were investigated, 2 found during the screening of 5,434 random group O donors, giving a frequency of 1:2717. The families showed the antigen to be inherited as a Mendelian autosomal dominant character segregating independently of: Rh, MNS and is not linked to X or Y. No pure examples of the corresponding antibody have been discovered in testing 6,580 sera from normal donors in South Wales.

Low density lipoprotein receptor activity in freshly isolated human blood monocytes and lymphocytes

Circulating human monocytes and lymphocytes were isolated by counterflow and density gradient centrifugation. Binding and degradation of low density lipoprotein (LDL) occurred predominantly in monocytes and to a much lesser extent in lymphocytes. The findings were consistent with greater LDL receptor activity in freshly isolated monocytes than lymphocytes, in keeping with differences in other cell surface receptors between these two cell types. Therefore, when freshly isolated mixed mononuclear cells are used to study LDL receptor activity in vivo in humans, careful attention needs to be given to the proportions of monocytes and lymphocytes, or alternatively, relatively pure preparations of monocytes should be used.

The significance of graft diameter

To study the influence of diameter on graft patency, an 8 mm aortoiiliac Dacron graft was implanted in on leg of 25 dogs that had liac arteries 3 to 5 mm in diameter and a 5 mm graft was placed in the other leg. In six dogs both grafts clotted within 3 months, in 10 dogs both grafts remained patient until the dogs were killed between 7 and 66 months, and in nine dogs one graft became occluded before the other. In eight of these nine animals the 8 mm graft became occluded before the 5 mm graft; only in one dog did the 5 mm graft become occluded first. When the 10 dogs with two patient grafts were killed, the 8 mm graft was found to be lined with thick, organized fibrin, whereas the 5 mm graft had a thin, smooth, glistening lining. Histologic examination confirmed that healing was more complete in the 5 mm graft. In vivo blood flow measurements in the dogs were used to compare flow rates and graft resistance in 4, 6, 8, 10, and 12 mm grafts. A given sized graft carried the same flow capacity as larger grafts until the flow rate was reached when graft resistance developed. Once resistance appeared, the graft could still triple or quadruple its flow capacity but it could not deliver the same rate of flow under the same pressure head as larger grafts. In 4 mm grafts, resistance first appeared at approximately 150 cc/min and capacity was 450 cc. In 6 mm grafts, resistance developed at 400 cc/min and capacity exceeded 1,200 cc. In 8 and 10 mm grafts, resistance was first noted at 800 and 1.400 cc/min, respectively. These studies suggest that 6 mm diameter grafts can carry the 200 cc/min or less that is measured in the human superficial femoral artery at rest, as well as the four-to sixfold increase that accompanies vigorous exercise.

The Redelberger antigen Rba

A new low frequency antigen, Rba, has been found in three blood donors. Studies on their families show that the antigen is inherited as a Mendelian autosomal dominant character. Rba segregates independently from ABO, MNSs P1, Rh, Kell, Duffy, Kidd, ACP1 and PGM1. Anti-Rba is not common in sera containing multiple antibodies ot low frequency antigens.

The effect of antibiotics, primary and secondary closure on clostridial contaminated open fracture wounds in rats

In a study of experimentally induced open tibial or femoral fractures in rats, after either closing or leaving the wounds open, the animals were given: no antibiotic, cephalothin (Keflin), or penicillin. The rats with wounds closed primarily and receiving no antibiotics had the highest mortality rate (11 of 25) from experimentally produced clostridial myonecrosis. The lowest overall mortality rate (5 of 99) was found in the penicillin-treated groups. The higher mortality rate in the femur fracture groups was probably because of the large muscle mass of the thigh. The importance of ideal anaerobic conditions for producing experimental clostridial myonecrosis is emphasized.

Endogenous immune complex nephropathy associated with malignancy I. Studies on the nature and immunopathogenic significance of glomerular bound antigen and antibody, isolation and characterization of tumor specific antigen and antibody and circulating immune complexes

Three patients with clear cell renal carcinoma and one with another intrarenal malignancy were studied for the presence of glomerular localized immunoglobulins, complement components and tumor specific antigen and antibody by immunofluorescence. To determine the association and elucidate the pathogenic mechanisms involved in the relationship between tumors and glomerular deposits, antibody eluted from tumor tissue and renal glomeruli, cryoproteins, serum antibodies and rabbit antisera to tumor tissue were tested for specificity to antigen. The relationship between tumor antigens and the lipoprotein antigen localized in normal proximal tubular brush border (RTE) and the small bowel mucosa, was studied by immunofluorescence, absorption and blocking studies as well as complement fixation. Immunoglobulins and complement components were localized in the glomeruli and tumor membrane of all patients. Sera and glomerular fixed antibody from three patients with renal cell carcinoma localized to normal proximal tubular brush border and jejunal mucosa as well as to tumor membrane and the glomeruli and proximal tubules of all of these three patients. Anti RTE activity was also detected by complement fixation. Immunologic similarity between RTE and renal cell carcinoma antigen was confirmed by absorption studies. Furthermore, cryoprecipitable complexes of tumor antigen and specific antibody were isolated from the serum. The tumor antibody was immunologically similar to RTE. In the other case the rabbit anti-tumor antibody and the patient's serum fixed to the tumor membrane and kidney of the patient but did not show cross reactivity with the renal cell carcinoma or RTE. These studies suggest that the tumor antigen in renal cell carcinoma is similar to RTE and the glomerular deposits represent tumor antigen and antibody complexes. In addition these investigations support the hypothesis that tumor immune complexes are significant in the glomerular lesions, and that the origin of renal cell carcinoma is in the proximal tubule. The investigations also show that tumor antibodies are specific for tumors of the same morphological type but not for other tumors in the same tissue. Moreover, the renal glomerulus appears to be a chosen anatomic site for deposition of tumor antigens and antibodies and studies of the kidney may provide insight into the nature of tumor antigens and antibodies. Cryoprecipitation appears to be a valuable method in isolation of tumor complexes and characterization of tumor specific antigen and antibody.