A Lethal Change in the Weather: Temperature Extremes and Premature Mortality

Extremely hot and cold days can be fatal to certain vulnerable populations, as the more than 160 deaths in two weeks during California’s July 2006 heat wave clearly showed. The elderly and lower-income individuals are generally acknowledged to be most vulnerable to the effects of temperature extremes, but relatively little is known about how such extremes combine with underlying medical conditions to increase mortality risks. Now, a team from the Harvard School of Public Health has analyzed millions of death records from 50 U.S. cities to identify factors that increase the risk of dying on extremely hot or cold days [EHP 114:1331–1336; Medina-Ramon et al.]. The study is the first of its size to identify specific diseases that produce the largest relative mortality increases on extreme temperature days. 
 
The researchers examined approximately 7.8 million death records for the period 1989 through 2000. They defined extreme temperatures for each city as the coldest 1% of daytime highs and the warmest 1% of nighttime lows. These are the most physically challenging conditions, with people unable to warm up even in the daytime or cool off even at night. The data were analyzed using a case-only approach, a technique borrowed from genetic research that allows the identification of time-invariant factors (such as gender) that modify the effect of a time-variant risk factor (such as weather). This approach allowed the researchers to compare the individual characteristics of those dying under extreme weather conditions with those dying on other days. 
 
The study’s large sample size provided enough statistical power for researchers to see how a variety of individual characteristics, including presence of chronic conditions, affected vulnerability to weather extremes for a specific cause of death. For example, previous studies had already shown that blacks are more likely than whites to die on a hot day, but the authors found that susceptibility in this subgroup was more pronounced when death was due to cardiovascular disease. Conversely, the elderly and diabetics were more vulnerable to heat when the primary cause of death was not due to cardiovascular disease. The researchers also found a large increase in vulnerability to heat in individuals who suffered from atrial fibrillation, a finding that has not previously been reported. Cardiovascular deaths, especially cardiac arrest deaths, also showed a greater relative increase on extremely cold days. 
 
The authors note that public health officials can target appropriate health services and infrastructure by knowing which subpopulations are particularly at risk from temperature extremes as well as the most common mortality causes that may affect them. This kind of information will be even more important as some subgroups (such as diabetics and the elderly) increase as a percentage of the population at large, and as global warming raises the probability of higher maximum temperatures.

Deletions of parts of the long arm of chromosome 6 have been reported in breast carcinoma (Devilee et al., 1991), ovarian carcinoma (Lee et al., 1990;Sato et al., 1990;Zheng et al., 1991;Saito et al., 1992a;Cliby et al., 1993;Foulkes et al., 1993) B-cell Non-Hodgkin's lymphoma (B-NHL) (Gaidano et al., 1992), T-cell acute lymphocytic leukaemia (T-ALL) (Menasce et al., 1994) and malignant melanoma (Trent et al., 1989). These results suggest the presence of one or more tumour-suppressor genes on chromosome 6q, an idea supported by chromosome-mediated transfer experiments (Trent et al., 1990). Devilee et al. (1991) reported a combined allelic imbalance (Al) at D6S37 (6q26-q27) and MYB (6q23.3-q24) of 50% in 42 informative patients. Furthermore, 90% of the allelic imbalance was shown to be due to allele loss rather than gain. Because these authors described 6q as 'the second most frequent site after 17p for LOH in breast cancer', we have carried out a detailed analysis of Al in malignant breast tumours from 42 patients using 18 recently described highly polymorphic dinucleotide repeat sequences (Weber and May, 1989;Weber, 1990) distributed along the length of 6q.

Materal and meod
Patients and tissues Blood and tumour samples were collected from 42 unrelated breast cancer patients with full clinical details. Their diagnoses were infiltrating ductal in 34 patients, infiltrating lobular in six, mucoid in one and indeterminate in one case.
Preparation of high molecular ,eight DNA The isolation of genomic DNA from blood lymphocytes and breast tumour tissues was performed as previously described (Sambrook et al., 1989 Six markers (D6S239, D6S246, D6S330, D6S355, D6S357 and D6S359) were isolated and characterised in our laboratory (Orphanos et al., 1993(Orphanos et al., , 1994; nine markers (D6S261, D6S280, D6S281, D6S283, D6S284, D6S286, D6S297, D6S300 and D6S313) were genetically mapped by others (Weissenbach et al., 1992); D6S220 was previously assigned to human chromosome 6 (Hudson et al., 1992). (CA), microsatellites for D6S186 and D6S193 loci, at 6q26 and 6q27 respectively (Saito et al., 1992b), were isolated from cosmids generously provided by Dr Y Nakamura using the method of Santibanez-Koref et al. (1993). Primers, polymerase chain reaction (PCR) conditions and physical mapping of the loci have been reported by Orphanos et al.  Table I. Allele analtsis Blood and tumour DNA samples from 42 patients with breast cancer were amplified by PCR to give radiolabelled products and analysed as described previously (Orphanos et al., 1993). Quantification of the autoradiographic signals was performed with a 425S phosphorimager and Image Quant software supplied by Molecular Dynamics. The ratio of allele intensities in the tumour sample (T) relative to that in lymphocyte DNA (B) was calculated from T = Tl/T2 and B = Bl/B2, where TI and T2 and Bi and B2 correspond to the intensities of alleles I and 2 in the tumour and the blood samples respectively. Factors of T/B <0.5 or T/B >2.0 were arbitrarily considered to respresent Al evidence for LOH. Samples that showed Al ratios close to 0.5 or 2.0 were either rerun on a denaturating gel or their analysis was repeated using fresh blood/tumour sample aliquots.  (Table II). Examples of Al are shown in Figure 1 for four informative patients. At D6S239 patient 42 shows no Al, whereas the tumour sample of patient 41 clearly shows an excess of the lower allele. Similarly, at D6S297 patient 33 shows no Al but patient 32 has relatively more of the upper allele in the tumour sample than in the blood.
The lowest frequencies of Al occurred at 6q13 in D6S313 (10%) and over five loci in the region 6q16.3-q25.2 from D6S283 to D6S355, where Al average 17.4% (Table II and Figure 2). Between these two sets of loci was observed the most frequent Al (mean = 35.3%) in a region that contained seven loci between D6S280 at 6ql3 and D6S286 at 6ql6.3-q21. Marginally increased Al was observed with D6S220 (26%) at 6q25.2-q27 and D6S193 (27.5%) at 6q27.
We amplified all 18 markers from the blood/tumour DNA pairs of 20 patients and 17 markers from another six patients. From the data shown in Table III it is evident that in many patients Al was very extensive. Thus, patient 32 showed Al of all markers in the distal half of 6q. Patient 31 showed Al of all informative markers except D6S239 and D6S359, which suggests that in some cases extensive Al may be the result of deletions involving more than one chromosome breakage and reunion event. In other patients, namely patients 5, 16, 24, 25 and 28, no Al was seen, and only single-marker A! was seen in patients 13, 18 and 39. Using this subset of 26 patients we compared the observed Al on 6q with the clinical parameters of histological site, staging, node involvement and oestrogen receptor status.
Since 92% of the tumours in the subset were of ductal onrgin compared with 81% in the total, it was not surprising to observe that the percentage Al values for loci in the subset reflected those seen within the data for all 42 patients samples. On the other hand, both lobular tumours (nos. 7 and 32) for which complete data were obtained showed Al at D6S261 and D6S359, distal to the major region of imbalance in ductal tumours. The small sample sizes did not allow us to determine the statistical significance of these observations and no other trends were observed.
The markers used in this study were regionally localised by hybrid cell deletion mapping (Orphanos et al., 1994), fluorescence in situ hybridisation (FISH) (Menasce et al., 1994) and by genetic linkage (Weissenbach et al., 1992). The order of markers given in this paper is that of the current consensus map of 6q (Volz et al., 1994) for markers from D6S313 to D6S261 (cen to q21 -q23). Markers distal to D6S261 are not on the consensus map, but their order is consistent with the partial maps from the quoted sources. The data presented in this paper also permit ordering of some loci by minimising the number of obligate breaks required to produce deletion of alleles in tumours. For example, patient 25 showed Al of D6S261 and D6S359, implying that the likely order of the three loci assigned to q21 -q23.3 is (cen)-261-359-357-(tel).
Proximal 6q (6ql3-q21) was highlighted by Lu et al. (1993), who observed del(6q) in 10 of 22 (45%) of breast tumours karyotyped, and by Thompson et al. (1993), who found del(6q) in 4 of 28 (14%) of primary breast tumours. Our results identify the proximal region 6q13-21 as the major site of Al in 6q in malignant breast tumours. The small sample size did not allow us to determine the statistical significance of the low Al values of 10%     (%) Key: 0, no AI; X, AI; NI, not informative (homozygous); NJD, not determined. Bold entries maximiise regions of AI.
with the flanking markers D6S313 and D6S283 and 19% with D6S284. If subsequent analysis of larger numbers of patients can confirm this finding, then the presence of at least two tumour suppressors may be implied, consistent with there being a site of frequent translocations in melanoma at 6qll-ql3 (Trent et al., 1989) and deletions in acute lymphocytic leukaemia at 6ql6.3-q21 (Menasce et al., 1994). Distal 6q was implicated as a region of high LOH by Devilee et al. (1991), the combined LOH of D6S37 (6q27) and MYB (6q23-q24) being 50% in 42 informative patients. The practice of grouping loci together to obtain a combined frequency for Al or LOH (Takita et al., 1992;Dodson et al., 1993) is not necessarily informative because the more markers that are involved the greater will be the additive effect of random losses. Thus, in our study, 16 of 26 (62%) tumours showed Al at one or more of the nine markers in the region 6ql3-q21 (D6S313-D6S283), and 15 of 26 (58%) tumours showed Al at one or more of the nine markers in the distal region 6q21-q27 (D6S261-D6S281).
Region 6q27, reported to show high LOH in serous ovarian tumours (Saito et al., 1992a), appears to show only marginally raised LOH (27.5%) at D6S193 in our study, suggesting that the implied associated tumour suppressor may be of lesser importance in breast cancer than in ovarian carcinoma.